Reviewed: https://review.opendev.org/c/starlingx/config/+/794906 Committed: https://opendev.org/starlingx/config/commit/75758b37a5a23c8811355b67e2a430a1713cd85b Submitter: "Zuul (22348)" Branch: f/centos8 commit 9e420d9513e5fafb1df4d29567bc299a9e04d58d Author: Bin Qian Date: Mon May 31 14:45:52 2021 -0400 Add more logging to run docker login Add error log for running docker login. The new log could help identify docker login failure. Closes-Bug: 1930310 Change-Id: I8a709fb6665de8301fbe3022563499a92b2a0211 Signed-off-by: Bin Qian commit 31c77439d2cea590dfcca13cfa646522665f8686 Author: albailey Date: Fri May 28 13:42:42 2021 -0500 Fix controller-0 downgrade failing to kill ceph kill_ceph_storage_monitor tried to manipulate a pmon file that does not exist in an AIO-DX environment. We no longer invoke kill_ceph_storage_monitor in an AIO SX or DX env. This allows: "system host-downgrade controller-0" to proceed in an AIO-DX environment where that second controller (controller-0) was upgraded. Partial-Bug: 1929884 Signed-off-by: albailey Change-Id: I633853f75317736084feae96b5b849c601204c13 commit 0dc99eee608336fe01b58821ea404286371f1408 Author: albailey Date: Fri May 28 11:05:43 2021 -0500 Fix file permissions failure during duplex upgrade abort When issuing a downgrade for controller-0 in a duplex upgrade abort and rollback scenario, the downgrade command was failing because the sysinv API does not have root permissions to set a file flag. The fix is to use RPC so the conductor can create the flag and allow the downgrade for controller-0 to get further. Partial-Bug: 1929884 Signed-off-by: albailey Change-Id: I913bcad73309fe887a12cbb016a518da93327947 commit 7ef3724dad173754e40b45538b1cc726a458cc1c Author: Chen, Haochuan Z Date: Tue May 25 16:16:29 2021 +0800 Fix bug rook-ceph provision with multi osd on one host Test case: 1, deploy simplex system 2, apply rook-ceph with below override value value.yaml cluster: storage: nodes: - name: controller-0 devices: - name: sdb - name: sdc 3, reboot Without this fix, only osd pod could launch successfully after boot as vg start with ceph could not correctly add in sysinv-database Closes-bug: 1929511 Change-Id: Ia5be599cd168d13d2aab7b5e5890376c3c8a0019 Signed-off-by: Chen, Haochuan Z commit 23505ba77d76114cf8a0bf833f9a5bcd05bc1dd1 Author: Angie Wang Date: Tue May 25 18:49:21 2021 -0400 Fix issue in partition data migration script The created partition dictonary partition_map is not an ordered dict so we need to sort it by its key - device node when iterating it to adjust the device nodes/paths for user created extra partitions to ensure the number of device node/path for each extra partition is calculated correctly, otherwise the adjustments could be messy that causes the partition DB update to fail. Tested AIO-SX upgrade with three additional partitions. Change-Id: I1cb3bbfaf144a59d29633c1784b0fde80529cd71 Closes-Bug: 1892554 Signed-off-by: Angie Wang commit 204d5c36438979767a5c30a31da332cfffde4e66 Author: Dan Voiculeasa Date: Mon May 24 10:45:43 2021 +0300 Expose ceph backend field over proxy endpoint The main storage backend endpoint acts like a proxy. It allows ceph backend to be configured, but doesn't return all the desired fields when doing a query. Fill the information about ceph backend network parameter. Story: 2008843 Task: 42350 Signed-off-by: Dan Voiculeasa Change-Id: I6521f3727ed96be33ade2419946c875d7ffb6e13 commit c6f0967086e61c7e55d8e0501f94168b82cb7040 Author: David Sullivan Date: Sat May 22 19:04:01 2021 -0500 Apply sriov/FEC configuration during SX upgrade During the SX we also need to apply the sriov/FEC manifest in order to configure the FEC devices. This needs to be done before the sysinv agent starts to maintain the host configuration. Closes-Bug: 1929301 Change-Id: I99bd891bf43fd5912d0297861fce819fe6ce678f Signed-off-by: David Sullivan commit 9b8a1d074f85dc3e246d164e5d0356e44906287b Author: Angie Wang Date: Thu May 20 22:26:46 2021 -0400 Ensure the old app plugins enabled in app recovery lifecycle When an application update fails, the app should be recovered to the pervious version. In the case that the new app version is decoupled(which has the plugins infrastructure) but the old app version is not, the old app version is not considered to be a system knowledgeable app which causes the plugins activation to be skipped during the subsequent app update so that an error occurs because the lifecycle operator is still looking for the new app version's plugins. Update to ensure the old app plugins are enabled before armada process during recovery and after recovery is completed. For the apps decoupled in stx5.0 but not decoupled in stx4.0 like nginx-ingress-controller and portieris, only reload the operators. The particular handling for non-decoupled apps can be removed in the release stx6.0 as all apps in stx5.0 are decoupled. Change-Id: Ief79baac428af7f926f8721f15ded340e3cf1e44 Closes-Bug: 1929149 Signed-off-by: Angie Wang commit 8423e70fd04f07bbf6a22eb83d45c719663b0c51 Author: Dan Voiculeasa Date: Fri May 21 12:56:47 2021 +0300 Fix pod max pids service parameter default value Openstack installation fails for rabbit-mq pods. Change the approach of how the default value is selected. Document recommended minimum values for apps instead of using them. Select the default value as high as possible, protecting against a rogue pod, protecting against platform slowdowns created by high number of processes in the system, but low enough such that platform is still responsive even on older hardware. User is free to decrease the limit to increase the degree of protection against slowdowns. Initially it was observed that openstack pods reach ~450 processes in steady state. New tests show even with the 2/3 extra room, 750 pid limit is not sufficient when deploying rabbit-mq pods. But 2000 is. Recommended minimum value for openstack pods pid limit becomes 2000. Partial-Bug: 1928949 Related-Bug: 1928353 Signed-off-by: Dan Voiculeasa Change-Id: I0d66173e2247fae15eda1ad0e83c7bcf858f0369 commit 7d740e60f41bf4ee57210caa7acddaa1e9c543fb Author: Don Penney Date: Thu May 20 23:04:47 2021 -0400 Add coredns SX to DX migration manifest As part of the SX to DX migration, a runtime manifest class is added to update coredns configuration. This commit updates the system modify handler to apply this runtime manifest when modifying system mode from simplex to duplex. Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/792494 Change-Id: I03b609de16d0273bf116f318affbe1f22867cfa9 Partial-Bug: 1929148 Signed-off-by: Don Penney commit 9d8cdc5bb33f9e8ae4dfd658a0c9b216e7557431 Author: Douglas Henrique Koerich Date: Fri May 14 14:45:29 2021 -0400 Replace applying flag by dict for FEC device check This change replaces the latest solution to check FEC device before an unlock action, that relied on an '/APPLYING' flag. In certain asynchro- nous scenarios, that flag could be cleared before than expected if an inventory report not related to the FEC device configuration came late (that might happen when configuring a long queue of SRIOV port changes) or by periodic sysinv report. The solution still uses the 'extra_info' field of PCI devices, this time "stringifying" a dictionary entry for 'expected_numvfs' that will keep (without clearing) at that field the programmed number of VFs at FEC device. It is then compared with the actual sriov_numvfs of device from the inventory report, in a similar way of what is currently done for comparing SRIOV interfaces (from database) to ports (from device). Closes-bug: 1927089 Signed-off-by: Douglas Henrique Koerich Change-Id: I380bd66a8229a72ef1981cbefa3a0543c28d7f30 commit 6d262e1b4c80ef6567edf7ebcaa494e143b30a77 Author: albailey Date: Thu May 20 08:43:15 2021 -0500 Fix zuul for bandit target Some zuul nodes running bionic do not consider the older version of pyflakes to be installable. This seems to be a cache issue. This fix updates the version of hacking defined in the top level test-requirements.txt file to use a more modern version. It also only imports yamllint if it is python 3, since the yamllint tox target is python 3 only. Partial-Bug: 1928978 Signed-off-by: albailey Change-Id: Ia7aa6a296810adc0d9ba9eca701ec70f2c4be8cd commit d55ca91b907c45dc73d4d44e39560e37a7b668ef Author: albailey Date: Wed May 19 13:51:15 2021 -0500 Specify the nodeset for zuul jobs The py2.7 jobs need to specify xenial The py3.6 jobs need to specify bionic The focal zuul nodes only have python 3.8 installed in them The copyright date was updated for some files in order to trigger the zuul jobs, as a no-delta type of change. Partial-Bug: 1928978 Signed-off-by: albailey Change-Id: Ifc7904d4908a5dbe2ffbd9214e5e4c425932afad commit f3edd0cd6a278d1cc537529ac366c23cc6d4919d Author: Lucas Cavalcante Date: Thu Apr 22 07:20:05 2021 -0300 Add pull_image option to perform_app_upload Downloading images at application upload enables subclouds to use the central cloud registry even when the central does not have the same application applied. This way Central would only need to upload an app such as Openstack in order to have images available for the subclouds. This is done by adding an extra option `-i` or `--images` that defaults to False. Testing: * VirtualBox: AIO-DX (Central), AIO-SX (Subcloud) * Upload an app to the Central Cloud using `--image` or '-i' * option (without applying it) Upload and apply app at * Subcloud using the DC registry Signed-off-by: Lucas Cavalcante Change-Id: I706c2bdf233617aadae8506724dde1afbbc1b35b Closes-Bug: 1925844 commit 30e0d90d24df92e79a414e33c1779aed5e43eb2c Author: Angie Wang Date: Wed May 19 10:48:44 2021 -0400 Update k8s application upgrade script for corner case The upgrade script may report the application update incorrectly because there is a window when querying the application version and status during update. In the case that the application update to the new version is failed, the query for the app version is made before the application recovery is triggered, and the query for the app status is made after the application is recovered to the previous version, the script gets incorrect information and reports the application is updated successfully. Update the application query in one request to eliminate this possiblity. Change-Id: Icbf173214de591861c21841e22359ad981453e84 Story: 2008055 Task: 42246 Signed-off-by: Angie Wang commit e4bc9dc6026f76964e91809d3971e73160d653db Author: Dan Voiculeasa Date: Mon May 17 13:23:54 2021 +0300 Fix skipping recovery of the old app for app updates When app update fails and it is requested that rollback is skipped it still reverts to the old version of the app. Confusion here was created by the names of the building blocks for the update logic (perform_app_rollback, perform_app_recover). In fact it is desired to skip operations that recover the old app. Added the missing logic for path of failed update operation. Now both upgrades and downgrades of the app behave the same. Tested by changing the pvc claim to trigger the armada failure. Closes-Bug: 1928671 Signed-off-by: Dan Voiculeasa Change-Id: I6792744257a9cb249e0b1bf99f9b78f3b27859d9 commit eec51d6a4a47737b1eba8585c412449292938ad9 Author: Vinicius Lopes da Silva Date: Thu Apr 29 10:55:40 2021 -0400 Delete ceph stor when removing controller When removing controller from cluster, its OSD is not being removed from ceph. This happens when removal is made by issuing "host-delete" command. This commit adds a call to remove the OSD from Ceph when controller is removed. Closes-Bug: #1926626 Signed-off-by: Vinicius Lopes da Silva Change-Id: If0fd9260ab7b9c717ef0a4ae621da0ffc9d0e6ab commit 32b4df542c4566ad08f057950f78dae7633a8233 Author: Jessica Castelino Date: Tue May 18 16:02:29 2021 -0400 Notify dcmanager when k8s upgrade completes When a k8s upgrade has been completed, we notify dcmanager so that it can do a kubernetes audit of the subclouds immediately rather than waiting up to an hour for the normal audit to run. Change-Id: Ife2abdbc65ad4ee91441db8fa39cb80291cbe201 Signed-off-by: Jessica Castelino Partial-Bug: 1928864 commit f9a862b5d80943b76cd71dab84c765f9b17af89e Author: albailey Date: Tue May 18 11:06:17 2021 -0500 Fixing pylint failures in zuul. pylint running in python3.8 will complain about some code such as 'import contextlib' Rather than destabilize the code, specify python3.6 for the pylint job. When python2 is fully dropped, the changes can be addressed. The zuul nodeset for bionic sets up a venv for python3.6 Closes-Bug: #1928841 Signed-off-by: albailey Change-Id: Ib2e7f467ce20e5b31aef7242405ef034583d4e1a commit bf547186d19a218c03e73f31c5d7cafd1a5d4bd3 Author: Dan Voiculeasa Date: Wed May 12 15:16:32 2021 +0300 Add service parameter to control pod pids limit Create a config section for kubernetes service. Create a parameter named pod_max_pids to have similar name as the kubernetes parameter pod-max-pids. Store the value in the config section. This will create a system-wide entry in hieradata when unlocking: plattform::kubernetes::params::k8s_pod_max_pids This affects hosts with kubelet running, meaning controller and worker personalities. A config out of date will be raised for all hosts of both personalities, even for parameters that target only a specific personality. After modifying the parameter a host-lock then host-unlock is required. Platform pods use under 20 processes in steady state. Some openstack pods reach ~450 processes in steady state. Since StarlingX provides some optional apps we provide a default value that takes into account the most hungry app, that being openstack. The database entry will be populated considering openstack will be applied.(I707ddc4ca67595fbf809c6ffc15ecd4fb21f4661), but we shouldn't restrict the minimum based on optional apps, as this allows the user to set a lower minimum if there is no plan to use openstack. Tested on Standard+dedicated storage: - out of sync raised for controllers and workers when using service-parameter modify - alarm cleared after host-lock, host-unlock - new value correctly generated and used - add with system service-parameter-add - modify with system service-parameter-modify Tested on top of: I10c1684fe3145e0a46b011f8e87f7a23557ddd4a Partial-Bug: 1928353 Signed-off-by: Dan Voiculeasa Change-Id: I74fcf2bd405c2a3811a4f27a55b28c0d001430e1 commit 82f01a8912567e48c0778689c80919e3fc6d9ce9 Author: Dan Voiculeasa Date: Mon May 17 18:44:29 2021 +0300 Re-add wrongly removed function in helm plugins Some app plugins that are based on base.BaseHelm use a function wrongly removed in I681ccb3302b8f233424bc291e08675a4dc2b10f7. Re-add the function. Closes-Bug: 1928696 Signed-off-by: Dan Voiculeasa Change-Id: I43fe2065714cbae2eebf43469424e812b8783218 commit cfe94d9dae6f55d5d44b23a57d742827f43b16ca Author: Dan Voiculeasa Date: Mon May 17 18:13:20 2021 +0300 Fix semantic check for N+1 app version Add the missing property 'relative_timing' for a LifecycleHookInfo object. Closes-Bug: 1928692 Signed-off-by: Dan Voiculeasa Change-Id: I5c87aac9497f19f3a5adb8d383465c78cdad0a7b commit 703b9dc6f97628271950f9b7352fdb8d4df8f74d Author: Pedro Henrique Linhares Date: Wed May 12 16:51:19 2021 -0300 Refactor and expose logic to acquire a flock with retries When locking the file descriptor skip_udev_partition_probe was not handling errors thrown by fcntl.flock which was leading controller-0 to degraded state after unlock. This change aims to strengthen that logic by handling the error properly, retrying the lock operation and improving logs. Re-implementation of commit cbb9121a289603ec003dec098b8fa5918ca98300. The original commit inadvertently replaced a shared lock with a exclusive lock on the decorator skip_udev_partition_probe which caused fd locking issues. This commit exposes utility functions to acquire shared or exclusive non-blocking locks of file descriptors. Tested on Standard (2 + 4) and AIO-Simplex configurations. Ran sanity load on both. Closes-Bug: 1922256 Change-Id: Ifcddab027df955152f420fd7451f42167694a31a Signed-off-by: Pedro Henrique Linhares commit d4f82539e0d421ab8a7f1cd466bdbc269727bddc Author: Teresa Ho Date: Thu May 13 22:23:59 2021 -0400 Leave parameter_reselect as null if not specified The sysinv API for interface returns the optional parameter 'primary_reselect' with its default value when the attribute is not specified. This update is to leave the parameter as null if it is not specified. Closes-Bug: 1928461 Change-Id: I67629aec1e58c26b1ed76c0cd1e37cd53e74b0b2 Signed-off-by: Teresa Ho commit ca1fbf08cbb2e0bdf3f98f7fa091af178768992d Author: John Kung Date: Thu May 13 14:00:12 2021 -0500 Update config file update operations to be more atomic sysinv agent iconfig_update_file() was unconditionally deleting the file or symlink to be updated prior to updating it. This could render the file missing if sysinv-agent is restarted. This update copies the updated file contents to the file and only removes symlink, if any, when needed. A few minor logging improvements related to logs observed in LP. Change-Id: I3c97e778b17dd0e9693a156ffe6b1c0269413f20 Closes-Bug: 1928368 Signed-off-by: John Kung commit b62a585ab191ac74e71ff6b4b57d716a3aa3ea2f Author: Angie Wang Date: Tue May 11 17:30:18 2021 -0400 Revert "Revert "Check for connectivity to the tiller postgres backend."" This reverts commit 46421279912daa62162e493c3455ae5b9b75cf69. One additional change made based on the orignal commit is calling retrieve_helm_v2_releases() instead of retrieve_helm_releases() to make helmv2 query only as only helmv2 is using postgres backend. Depends-On: https://review.opendev.org/c/starlingx/config/+/789828 Change-Id: Ia3c52192cea7c3addec446b22436db7a028ec5bc Signed-off-by: Angie Wang commit 12fff41d7803c7cea2b34e356ac65d361ca57789 Author: Andrei Grosu Date: Wed May 5 13:03:50 2021 +0000 Handle empty 'helm list' result when there is nothing deployed The existing code assumes that there are always applications deployed and the result is never an empty list. The previous implementation ignored the return code when the subprocess was killed by the timeout handler. Split the method in two submethods for helm v2 and v3 implementations. Closes-Bug: 1923587 Signed-off-by: Andrei Grosu Signed-off-by: Angie Wang Change-Id: Ib547bdb20c39e35c1538e3abb90108f7e3cad228 commit 858aee342dcb94654c9fcd8f0731a2838d845eb1 Author: Douglas Henrique Koerich Date: Thu May 6 07:35:56 2021 -0400 Check for FEC device configuration pending SRIOV configuration at Intel ACC100 device configuration was not being correctly applied when it followed a bunch of other configurations, like data networks, SRIOV interfaces, etc., and soon before the unlock of the host, because runtime manifest was not finishing before system shutdown. "This issue came about because of the way we handle 'pci_devices' in contrast to 'interfaces/ports'. For the interfaces/ports, when configuring SR-IOV, the interface sriov_numvfs will represent the user requested value, while the underlying port sriov_numvfs will represent the actual system value. Then, before a system can be unlocked, the values are compared for equality. This ensures that the runtime manifest that is setting the value of sriov_numvfs has a chance to run. For 'pci_devices' like the FPGA cards we support, there is no concept of an 'upper interface'. Therefore, the value of sriov_numvfs for a pci_device represents the system value rather than the user requested value. This can cause issues when performing and unlock right after configuring a pci_device with SR-IOV. There's a chance that the system populates its hieradata and unlocks before the runtime manifest has had a chance to configure the SR-IOV value for the pci_device." (Steven Webster) This change appends a "/APPLYING" string to 'extra_info' field of FEC device when it gets configured via API, and it will be removed only when the corresponding inventory is reported back by SYSINV agent. Until there, attempts of unlocking the host will find that sub-string at the field and this is subject of additional semantic check. Closes-Bug: 1927089 Signed-off-by: Douglas Henrique Koerich Change-Id: I175bc01a2a51808c4dc7b821905c7417660bf286 commit a6481bc4d169ab3c81c73b6807299d3febf7d591 Author: Andy Ning Date: Wed May 5 09:34:00 2021 -0400 Fix invalid admin endpoint cert during subcloud upgrade cert-mon queues failed cert update tasks and retry them later on. But the retry periodic function is not started in subcloud so it doesn't really work. This commit fix it by starting these periodic functions if the system's DC role is subcloud. This commit also added unauthorization exception handling for platform cert update, so that the retry task will reattempt updating the cert with a new token. The other certs update already have such exception handling. Note, commit 862c1746abb8d8901d2acb4bcb43569210e55f3e is needed to fully fix Bug 1926788. Closes-Bug: 1926788 Signed-off-by: Andy Ning Change-Id: If7f631ee3e5f97db7a06b184f9e68cf901cc8344 commit 4e75297b58c4a488451d575819608ec203c89972 Author: Andy Ning Date: Mon May 10 12:24:52 2021 -0400 Specify default connect_timeout for sysinv agent During bootstrap, for some reason rabbitMQ takes longer time to response to sysinv agent's AMQP request after TCP connection is established. This usually happens after multiple connection failures, and can be as long as 8s. But it's still within th 10s handshake_timeout configured on rabbitMQ. However the sysinv agent uses the default 5s timeout from kombu lib so it doesn't wait long enough before hanging up the connection. This causes the agent can't reconnect to the rabbitMQ. In turn, the defered manifests don't have chance to pass readiness checking and get applied. This commit specifies a 10s default connect_timeout for the agent t0 connect to rabbitMQ, aligning with the server side timeout. Closes-Bug: 1928008 Signed-off-by: Andy Ning Change-Id: I8cc476910f47fca687ddcfd5f3d20f451f70ffce commit 55e70b52d73c0c7948fa186a741600c190f0cd2a Author: Pedro Henrique Linhares Date: Mon May 10 12:27:04 2021 -0300 Fix bootstrap error on AIO-SX due to post SX to DX migration actions During start-up of sysinv-conductor we test if the AIO-SX to AIO-DX migration is occuring and it relies on an active ihost being available which is not the case for when the host is bootstrapping. This commit adds a check to whether host_uuid and ihost are available before usage. Closes-Bug: 1927984 Signed-off-by: Pedro Henrique Linhares Change-Id: I3287852242579164604a7815a599c4f7e9f704f8 commit 6df2034a4e9e2a25cd0ba39af7074fec5a26466d Author: Pedro Henrique Linhares Date: Wed May 5 11:34:47 2021 -0300 Adding AIO-SX to AIO-DX migration steps patching existing PVs Kubelet and kube-api are no longer available during puppet manifest run during unlock. Therefore, we moved the patching of Persistent Volumes from puppet tosysinv-conductor as a post-migration step during its start-up. Closes-Bug: 1927224 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/789844 Depends-On: https://review.opendev.org/c/starlingx/fault/+/790183 Change-Id: I9745b7f8547c82485353130156011650f2655317 Signed-off-by: Pedro Henrique Linhares commit 407a3e374815c7d325b650e62205e4898ab89b13 Author: Rafael Jordão Jardim Date: Fri Apr 23 13:26:49 2021 -0400 Send the binary data instead of path The issue is when it executes a system application-upload in the remote cli it returns an error "Application-upload rejected: application tar file /wd/custom_apps/hello-kitty.tgz does not exist" the reason is because the cgts-client sends the path of the tarbal, the proposal solution sends the binary data and save it in a path on controller Closes-bug: 1926308 Signed-off-by: Rafael Jordão Jardim Change-Id: I6dadef9e86612328ae68fc90564e929646e93dba commit 9e9979d96e99c7cf8484f28a93d0635f3012fc5d Author: Jessica Castelino Date: Thu May 6 15:30:43 2021 -0400 Specify timeout for _get_token REST API request in cert-mon In a DC env, when management network connectivity is disrupted for some time, cert-mon's audit greenthreads hang. This happens because of missing timeouts in REST API requests which cause the urlopen to remain hanging forever. As a result, the dc-cert sync status for the subcloud remains unknown indefinitely and the subclouds remain out-of-sync. This commit adds a timeout to the _get_token method to avoid this issue. Change-Id: Idd41cfca6b28287de8328b1ace856cc391778cac Signed-off-by: Jessica Castelino Closes-Bug: 1927735 commit 62c9e55efd1bc1f713924d3d08d7ba42e3a94af0 Author: Charles Short Date: Tue Mar 23 11:31:21 2021 -0400 Switch type(certificate) and type(obj) Switch type(certificate) and type(obj) to instance since its accords with python style. Story: 2006796 Task: 42442 Test: - Ran unit tests. - Built iso and ran some smoke tests. - Tested on general install on controller+storage+workers system, bootstrapped ok. Signed-off-by: Charles Short Change-Id: I03d444f3ae67ab802881b8ede90f1e1c7e9694cc commit ddc6b69dfc82487fc22f3c2eb1795d9bfdb2e155 Author: Dan Voiculeasa Date: Tue Apr 20 12:16:17 2021 +0300 Allow configurable ceph storage backend network Ceph components are configured by puppet. Hieradata for it is generated by sysinv. Currently the configuration is generated using IPs from the management network. Allow ceph components to be configurable on cluster-host network. Create a 'network' parameter for strorage-backend-add command. The parameter is optional, the rest controller will default the value to management network. Example usage: system storage-backend-add --network cluster-host ceph --confirmed system storage-backend-add --network mgmt ceph --confirmed system storage-backend-add ceph --confirmed Updated unit tests. Added unit test for component that generates the ceph monitor ips for hieradata. Tests: 1) AIO-SX: add ceph using --network cluster-host, unlock, ip is from cluster-host network. 2) STANDARD 2+2: full deploy, no --network parameter, everything as before. 3) storage-backend-modify and storage-backend-add allows only 'mgmt' and 'cluster-host' network parameter. 4) storage-backend-modify does modify the network before the first unlock; is rejected for network parameter after first unlock. Story: 2008843 Task: 42350 Task: 42351 Signed-off-by: Dan Voiculeasa Change-Id: I681ccb3302b8f233424bc291e08675a4dc2b10f7 commit 6f15428e7565736ab600f4f0f01e1445987cc3a6 Author: David Sullivan Date: Thu May 6 10:12:44 2021 -0500 Add logging to secured-etcd upgrade script Add logs to 70-active-secured-etcd-after-upgrade.sh. Redirect ansible logs to a named file under /root. Closes-Bug: 1927511 Signed-off-by: David Sullivan Change-Id: I86fd964cad02ebb8eeb8645fc9dea71a3251aef7 commit 37d348ae6db2bf8522737287d805a100ed8245e5 Author: David Sullivan Date: Thu May 6 10:02:50 2021 -0500 Limit etcd migration to swact actions Currently the etcd migration will take place whenever upgrade_swact_migration.py is called. Typically this is called during a swact as part of etcd start. However the script can also be called during etcd restart, which occurs as part of activating secured etcd. The solution is to limit the etcd migration to controller-0. This will ensure the action will only be taken during the swact to controller-0. Closes-Bug: 1927508 Signed-off-by: David Sullivan Change-Id: I893eeaa7ddb0b600baa051498a30ed737c688151 commit 3e9982bab088df428e786d5fdf2515dc21f189fa Author: Andre Fernando Zanella Kantek Date: Wed May 5 17:21:35 2021 -0400 AIO-SX reboots after change OAM ip address On HW tests, it was detected that openstack-endpoints restart was happening at the same as the service-manager restart, this creating a conflict that preventing SM services to reach enabled-active. This was provoking the reboot. The correction add a class to execute openstack-endpoint runtime restart on the post stage on puppet, avoiding to be run as SM is restarted. Tested on AIO-SX, by monitoring manifest apply and validating that no reboot happens Closes-Bug: 1927275 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/789946 Signed-off-by: Andre Fernando Zanella Kantek Change-Id: I9f547fbcc73ba5fea077c764a4a9282a02ac71c6 commit c43772ca5c8519254d2be4b16c3326a90091f806 Author: Adriano Oliveira Date: Wed May 5 00:36:59 2021 -0400 Memory adjustment to consider memory available In order to handle patch applications that might change the reserved constants defined for the platform memory, there is an auto adjustment routine that compensates the potential extra memory by decreasing the number of hugepages configured. However, the routine should take into account the memory available and only decrease in case a certain threshold is reached. For the sake of safety on memory allocation, it was considered 50% of total memory to be enough safe guard to avoid reaching the total memory limit as well avoiding hugepages to be removed (mainly 1G ones). Also, vswitch hugepages allocation have been added to the calculation of memory allocation. Testing: 1. Patch that increases memory allocation constant 2. AIO-DX upgrade from stx 4.0 to 5.0 (in which the platform memory reserved increased from 7000 MB to 8000MB) Closes-Bug: 1927172 Signed-off-by: Adriano Oliveira Change-Id: I0bb29de83268709cdae07834f537a28010d884bc commit 6ac2c0e3d0e74dd8d139dba06afe96b6f438955e Author: Daniel Safta Date: Tue Apr 27 10:36:44 2021 +0000 check app progress before swact This update adds a new check which will reject a swact while an application apply is in progress. Closes-Bug: 1926405 Change-Id: I3c683776f3ecaf9c78d111b5b1108e9582497aaa Signed-off-by: Daniel Safta commit c9df464849cfe9f6ee769d1682f0968a7aa7af6f Author: Teresa Ho Date: Tue May 4 09:15:28 2021 -0400 Fix missing sriov attributes in N3000 pci info The fpga-agent did not include all of the sriov attributes in the reporting of N3000 pci device info to the conductor resulting in fields being overwritten. This update ensures that the attributes sriov_vf_driver and sriov_vf_pdevice_id are copied. Closes-Bug: 1925513 Change-Id: I891166b1a0966b3bd9bfe253f954ff22f7ea8677 Signed-off-by: Teresa Ho commit 862c1746abb8d8901d2acb4bcb43569210e55f3e Author: Bin Qian Date: Fri Apr 30 12:14:31 2021 -0400 Remove subcloud admin endpoint data migration Admin endpoint cert upgrade will be handeled by manifest, so data migration is no longer needed in subcloud. On N+1 side, admin endpoint cert secret (key/cert) will be pulled directly from k8s resource for manifest to generate endpoint cert on first host unlock. Only need to update SAN of admin endpoint cert. Closes-Bug: 1923510 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/786666 Change-Id: I4312abd6c767d6ba54c13ce1e90f2e25df9ed216 Signed-off-by: Bin Qian commit 67c214a4e4ac234ceab33d1514f550eea41ae928 Author: John Kung Date: Thu Apr 29 07:55:55 2021 -0500 Clear host config_target on upgrade migration The host config_target is taken as a snapshot on the upgrade-start. This can lead to config out of date condition if the N controller issues subsequent config requests. This is more likely in duplex controllers, however, for consistency the config_target is reset to track when the N+1 controller is active. Clear the config target of all hosts on upgrade migration. The N+1 controller will resume tracking the config by generating config_target when configuration is issued. Condition for upgrade activation-complete is updated to account for potential None config_target. Verified upgrade complete on AIO-SX and Duplex systems. Change-Id: I4dd44e6548a45d32ab6a0b6735a04d624da7caad Closes-Bug: 1926512 Signed-off-by: John Kung commit 46421279912daa62162e493c3455ae5b9b75cf69 Author: Angie Wang Date: Sun May 2 21:41:04 2021 +0000 Revert "Check for connectivity to the tiller postgres backend." This reverts commit 5edd3bdbe588e2c2e7a58cb839f030305613c30f. Reason for revert: It causes ansible bootstrap failed. Change-Id: I5f3640db8576eab1500b54f111a06a981c98b599 commit 5edd3bdbe588e2c2e7a58cb839f030305613c30f Author: Andrei Grosu Date: Tue Apr 13 08:52:40 2021 +0000 Check for connectivity to the tiller postgres backend. The existing code checks that the pod(s) are 'Running' but that might not be enough as the service inside the pod (postgres) might not be able to accept connections. Closes-Bug: 1923587 Signed-off-by: Andrei Grosu Change-Id: Ide49e4a38b805d5fc41d9f06d94393c69c6ed9d2 commit d225a0217bf3dae22c12e1266c4beab78f1032dd Author: David Sullivan Date: Sun Mar 28 14:02:32 2021 -0500 Run SRIOV manifest during controller-1 migration To address another issue we blocked manifests from applying on a controller with a different version from the active controller. This blocks the sriov manifest from running on controller-1 after it has been upgraded to N+1. To address this the data migration will check if controller-1 has sriov interfaces configured and apply the manifest if necessary. The previous change: https://review.opendev.org/c/starlingx/config/+/766637 Closes-Bug: 1921788 Change-Id: I1ec790fced00a4e973e546a260d28f52ef06fb3a Signed-off-by: David Sullivan commit e53cf0cc2cb4490a93405789faf8fe497d8894ba Author: Melissa Wang Date: Fri Apr 23 16:47:43 2021 -0400 SX-to-DX: Check host administrative state This update adds semantic checks to ensure that the controller is locked before starting simplex-to-duplex migration. Story: 2008587 Task: 42369 Change-Id: I1ebf3bb531073344b4d17a00cd1ee480051f3897 Signed-off-by: Melissa Wang commit b9130523851ef8f8f2f300d4b583c76c012c51d4 Author: Yuxing Jiang Date: Tue Apr 20 10:37:47 2021 -0400 Add RPCAPI calls to apply LDAP client and DNS runtime manifest This commit adds RPCAPI calls to invoke runtime manifest of LDAP client and DNS after adding system controller network and system controller OAM network in a subcloud after the initial network configuration completed. Tested: Delete the system controller network and system controller OAM network in a subcloud and add them with different values. The related configuration and Hieradata are updated. Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/787750 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/785977 Story: 2008774 Task: 42307 Change-Id: I4ddf88efa16299c9415f4bf156f2be57e8cc826e Signed-off-by: Yuxing Jiang commit a9743b3e05c61a0b738af4770eeff5727bc51622 Author: Andre Fernando Zanella Kantek Date: Tue Apr 27 07:47:21 2021 -0400 Change in oam ip fails to update the system endpoints, in AIO-SX With story 2008531 the OAM-IP update is not changing the IP addresses for the openstack endpoints, although the REST APIs are already responding with the new IP The correction adds the keystone's endpoints class to the puppet manifest class list in order to correct the situation Tested in AIO-SX setup, as described on the launchpad Closes-Bug: 1926288 Signed-off-by: Andre Fernando Zanella Kantek Change-Id: I57c4a3392879be3c94de89ee67af3c5a50095b54 commit a9fc0be4e500a8739db04b7555c1be50a877b21f Author: Yuxing Jiang Date: Fri Apr 16 08:19:25 2021 -0400 Support configuration of system-controller address pools post-install During migrating a subcloud to a new central cloud, the system controller's network configuration needs to be re-configured in the subcloud. This commit removes the check of is_initial_config_complete flag during modification/deletion of the system-controller-subnet and system-controller-oam-subnet address pools. After this commit, the two types of address pools can be re-configured after the initial bootstrap. Test: Delete the system-controller-subnet and system-controller-oam-subnet in a subcloud after bootstrap. Change-Id: Ied68bbfd83a0cc1c3bb0fb31ee55f924353fb4b5 Story: 2008774 Task: 42291 Signed-off-by: Yuxing Jiang commit 754c6861ca33a581f3a32a7a353ec9672dd0c8b9 Author: Charles Short Date: Mon Apr 26 11:06:45 2021 -0400 Fix zuul errors due to changes in dependencies Pin hacking to < 4.0.1 to fix zuul gate issues. Test: Ran tox -e pep8 command to validate the flake8 job and result. Related-Bug: 1926172 Signed-off-by: Charles Short Change-Id: I74fec1c352b8947b58498e32b7554e54c77aaeaa commit 38ae08893bcb76142da1c1b1133a29f8a767bf16 Author: Sabeel Ansari Date: Thu Apr 8 14:13:28 2021 -0400 Verify cert chain after adminep update This commit checks the validity of the certificate chain after admin-ep certificate is renewed on subclouds. If the check fails, the update_admin_ep_cert deletes the secret and subsequent audits will recreate. This check fixes the scenario where old ICA was used when updating & renewing rootCA. Closes-bug: #1923071 Signed-off-by: Sabeel Ansari Change-Id: I112dd52e220dcc8bdb72c2c772ede72fbb786c7b commit 3c90e3b949c5533feee25efedda114376f43a0ff Author: jgauld Date: Mon Apr 19 17:21:09 2021 +0000 Add lifecycle hook to allow to_app application-update semantic checking This adds new lifecycle update operation so that a semantic check lifecycle step is run against the "to" application during application-update. The application specific lifecycle code requires a custom semantic check using the new hook (e.g., similar to the following): if hook_info.lifecycle_type == constants.APP_LIFECYCLE_TYPE_SEMANTIC_CHECK: if hook_info.operation == constants.APP_UPDATE_OP: if hook_info[LifecycleConstants.EXTRA].get(LifecycleConstants.TO_APP, False): return self.update_check(app_op._dbapi, app) Testing: * VirtualBox: AIO-SX * Created example application changes for platform-integ-apps using semantic check shown above, and prototype update_check() routine. * Tested the following for both pass and fail semantic check cases: system application-update platform-integ-apps-1.0-29.tgz * Tested that apps on the application-update are okay on an upgrade, as invoked in upgrade-scripts/65-k8s-app-upgrade.sh. i.e., system upgrade-activate Semantic check passes, update proceeds to 1.0-29 "to" version, the 1.0-27 "from" version is cleaned up. Semantic check fails, application recovery proceeds to "from" version, the 1.0-29 "to" version is cleaned up. Example when semantic check fails, system application-list | platform-integ-apps | 1.0-27 | platform-integration-manifest | manifest.yaml \ | applied | Application update from version 1.0-27 to version 1.0-29 aborted. \ Application recover to version 1.0-27 completed. Please check logs for details sysinv.log sysinv 2021-04-19 09:32:53.422 3810926 INFO sysinv.conductor.kube_app [-] \ Starting recover Application platform-integ-apps from version: 1.0-29 to version: 1.0-27 Story: 2008829 Task: 42311 Signed-off-by: jgauld Change-Id: If0787e3e3806bdf5dc175fde64ac63e1f38fd852 commit 251afbcdfb5a3f050921275bca97cb9de9553572 Author: Jessica Castelino Date: Mon Apr 12 14:51:50 2021 -0400 Handle "updating" and "recovering" app update states during upgrade activation During simplex subcloud upgrade, if the initial upgrade activation request fails, dcmanager orchestrator re-tries the activation by sending another request (current logic is to retry up to 10 times). On the second upgrade activation request, sysinv skips the oidc-auth-apps reapply as the app is in 'updating' state and proceeds to completing the remaining steps of the activation sequence. As a result, upgrade activation completes while platform apps were either in the incorrect version or incorrect state. This commit resolves the issue by skipping only for "uploaded" and "applied" states during upgrade activation. Change-Id: I4b0aa4897e83a47ccdcf58c37232301f3668de32 Signed-off-by: Jessica Castelino Story: 2008055 Task: 42246 commit b1c8d95f2cf8965eccf3306d372a8c820edd6634 Author: Mihnea Saracin Date: Tue Mar 30 14:12:23 2021 +0300 Fix resize of drbd filesystems This commit adds the following modifications: - The drbd filesystems sizes are now calculated using 'dumpe2fs' utilitary because it gives better results for larger filesystems. - Added extra checks before and after executing 'resize2fs'. Before running 'resize2fs' check if the drbd device is resized using "/sys/block/{drbd_device}/size" and sector size. After running 'resize2fs', check if the filesystem is resized using 'dumpe2fs'. - The drbd filesystems were resized only if they were in SyncSource or PausedSyncS states. There are cases when drbd-overview showed “Connected” instead of this sync states and the filesystems would never be resized. Now the drbd filesystems will also be resized when they are in "Connected" state. Closes-Bug: 1921896 Change-Id: I548300deb8916ce863bcd4bb70969cb9d51c9c2a Signed-off-by: Mihnea Saracin commit dd71459015c838dbd68ed90c5eb3ee16b5827004 Author: Pedro Henrique Linhares Date: Thu Apr 8 13:02:33 2021 -0300 Adding support for Ceph storage during Simplex to Duplex migration This change will allow users to migrate a simplex system with Ceph enabled to duplex via system modify command. During unlock it will get the cephfs filesystem pool names and generate the necessary hieradata. This file along with other puppet changes on stx-puppet will be used to perform necessary changes on the Ceph cluster to support a duplex configuration. Story: 2008587 Task: 42079 Signed-off-by: Pedro Linhares Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/783727 Change-Id: Idaa7ebbf3a9c55658187e1d5ca6c357349659d43 commit ca6a15adffc06cd4979015dbe596458fdf309f9b Author: Marcus Secato Date: Mon Apr 19 20:32:54 2021 +0000 Revert "Adjust lock acquiring logic" This reverts commit cbb9121a289603ec003dec098b8fa5918ca98300. Reason for revert: reverting as this caused issues in DC upgrade Change-Id: Ie665b9c4e4d1280d7c8a0821cb7995c9374ce02f commit a24cd707a5cafaa2d49b25d1d5285c5c8582eb5d Author: Angie Wang Date: Fri Apr 16 12:46:52 2021 -0400 AIO-DX: Controller-1 fails to be unlocked after downgrade During stx4.0 to stx5.0 upgrade, controller-1 fails to be unlocked after downgrade due to the incorrect disk partition and physical volume information stored in stx4.0 DB that causes the puppet manifest apply failed during unlock. This is because cgts-vg size is decreased in stx5.0 and after controller-1 is upgraded to stx5.0, additional partition and pv are created at stx5.0 side to match the size in stx4.0. However, controller-0 is still running stx4.0 DB and it gets updated with the new created partition and pv info sent from controller-1 sysinv agent audit. This commit updates to ignore the disk partition and physical volume information sent back from a different version during upgrade. Tested: - AIO-DX upgrade from stx4.0 to stx5.0, verified upgrade is completed - controller-1 downgrade after it is upgraded and unlocked, verified upgrade abort is completed Change-Id: I5d7858e4b29d096437a5ddf94cd78c74fadfacad Closes-Bug: 1924786 Signed-off-by: Angie Wang commit 0271808d75fc12909904e27103a1fd3455df3f18 Author: Charles Short Date: Wed Mar 17 11:10:52 2021 -0400 cgts: Add missing dependencies Prettytable and six are required dependencies that were not listed in the requirements.txt. Story: 2006796 Task: 42071 tests: - Ran build test. - Ran basic smoke test with resulting RPMS and iso. Signed-off-by: Charles Short Change-Id: Ib047e61487ed26792930c80db167920afc21c66c commit bc34618d83a5e38d765cbde0a91ebf384b8857b8 Author: Rafael Jordão Jardim Date: Tue Apr 6 07:44:06 2021 -0400 Python 2 to Python 3 compatibility The changes of https://review.opendev.org/c/starlingx/config/+/782575 was used to test the cgts-client, so it is need to be merged first. Removing python-neutronclient because this dependency is unnecessary, it was removed by copying a few very small utility functions from python-neutronclient into the cgtsclient. Development: When I was trying to find things to modify I followed the approach of build the client, get the tar file, I set up 2 environments one based on python2 and another python3, I installed the tar client in both environments and i exported the env vars that the client expect to get to request the controller, and doing that I could switch between the two python and indentifying what I should modify. Test: After all the modification I built an ISO and I installed that to run some commands and check if my changes got any side effects. After that followed the procedure to update the remote CLI docker image and insert the updated client there and I test this new image in the remote CLI. Story: 2007106 Task: 42268 Depends-On: I5086832605752bdb00a40a24596494c8fd987692 Signed-off-by: Rafael Jardim Change-Id: Ibf919260693f1cbe99993d1de01ecf785d604839 commit 8cc522bed57aac4699fed1b2cdf7cf4b213a4dc7 Author: Adriano Oliveira Date: Tue Apr 13 19:21:54 2021 -0400 sysinv-api script to return ERROR Remove NOT_RUNNING return in case sysinv-api pid is still active but fails to ping. This scenario can happen in case sysinv-api is manually killed and SM triggers a start. Also, on the new routine to check if sysinv-api is properly replying, return ERROR instead of NOT_RUNNING for SM consistency. Testing: Tested double sysinv-api kill withing 90 seconds as per SM configuration. Upon first sysinv-api kill it should be restarted, upon second sysinv-api kill, if done within 90 seconds, a swact is triggered. Also verified that sysinv-api request routine was engaged. Also tested AIO-DX boostrap, manual swact via host-swact and patch application. After boostrap and patch application tested double sysinv-api kill and manual swact. Other fix related to this issue was addressed by this change: https://review.opendev.org/c/starlingx/stx-puppet/+/783980 Closes-Bug: 1893669 Signed-off-by: Adriano Oliveira Change-Id: I1b1ab0560237f602dadf074331f6a165d12330c7 commit 7ce3d16eeaa4c186e70aad56a1c92a8279dd0aae Author: Bin Qian Date: Thu Apr 8 11:08:27 2021 -0400 Add sysinv-reset-n3000-fpgas cmd When AIO runs single manifest, reset N3000 FPGA needs to complete without docker local registry and other SM managed services. This adds sysinv-reset-n3000-fpgas cmd for puppet to reset N3000 FPGAS at host start-up. The sysinv-reset-n3000-fpgas cmd separates the function of reseting n3000 fpgas from sysinv-fpgas-agent as sysinv-fpgas-agent has dependency to rabbit, which is not available until manifest completes. Change-Id: Ic3c4b2a00515d194793257729362f71e2951286c Partial-Bug: 1918139 Signed-off-by: Bin Qian commit 6acd2e3564d3d708e496c1a7e78b064419f1fdbf Author: Bin Qian Date: Tue Feb 23 12:59:28 2021 -0500 Single puppet manifest for AIO controllers Create a single puppet manifest for AIO controllers. This change includes: 1. remove workerconfig from an AIO controller deployment 2. running puppet based on subfunctions of the nodes Depends-on: https://review.opendev.org/c/starlingx/stx-puppet/+/780600 Partial-Bug: 1918139 Signed-off-by: Bin Qian Change-Id: Ie3693219e3c19460ac5b617cc216cbc809ec2403 commit ad8567f06485a10edf3857fbc87ae7d3058a1dfc Author: Gustavo Santos Date: Tue Apr 13 16:09:21 2021 -0300 Restart tiller on openstack pending install check This is another attempt at fixing the same bug as the merged review https://review.opendev.org/c/starlingx/config/+/783472 had tried, since there were reports indicating that the bug would still occur on certain setups. This patch explicitly forces a tiller restart when catching the first HelmTillerFailure exception caused by the broken pipe error, instead of only trying to rerun the 'helm list' command, which was believed to be a reliable workaround to the problem, but didn't solve it in every possible scenario. Closes-Bug: #1917308 Signed-off-by: Gustavo Santos Change-Id: I38667609173ca5c6fed028f75742ae99efedf149 commit 5a9582ac5ca84db7dfaaa2474f05d373224f3d63 Author: jgauld Date: Tue Mar 16 06:40:49 2021 +0000 Platform support for application upgrades This enhances the application management component of sysinv to support the upgrade of both platform and non-platform applications. This supports new application metadata: upgrades: update_failure_no_rollback: from_versions: - - supported_k8s_version: minimum: maximum: supported_releases: : - - ... The processing of patch_dependencies is deprecated in favour of supported_releases. This enables applications to specify: * required patches such as a platform patch_level which the application can run on for a specfic release * supported K8s Versions which the application can run on * app releases that this version can upgrade from * application update failures do not rollback Application upload and updates check against this metadata. System health pre-upgrade checks make sure that: * all applications are in a valid state for the upgrade (i.e. uploaded or applied) * active controller is controller-0 Testing: - Configs: AIO-SX, AIO-DX, Standard - new metadata format validation with valid/invalid input - upversion platform-integ-apps tarball with new metadata, and do an application-update e.g., system application-update platform-integ-apps-1.0-22.tgz * applies if the new metadata not provided * blocks update if from_versions, supported_k8s_releases, or supported_releases criteria not met - normal operation of system application-remove/application-delete - system application-upload should apply/block based on metadata - trigger 'no_rollback' handling by forcing an application rollback using a small manifest.yaml timeout and too many replicas that cannot be deployed - AIO-DX after load-import, swact to controller-1, verify that upgrade is blocked - perform upgrade of current N to N+1 release * upversion platform-integ-apps in N+1 release with new metadata * import new load containing new metadata * system health-query-upgrade; should indicate fail and failure reasons based on new criteria * follow 'system upgrade' procedure; should block upgrade based on new criteria Story: 2008055 Task: 42179 Signed-off-by: jgauld Change-Id: I93a6f2ada7ce52414190948cbc458f6295c50603 commit d1a13c9b3f3458e9744cd5eed675157befa45c77 Author: Gerry Kopec Date: Thu Apr 8 19:48:13 2021 -0400 Increase platform memory reserve on AIO hosts Increase memory reserve for AIO by 1000 MiB. Total memory reserve for AIO in MiB will be 5000 + 2000 + 1000 * number of numa nodes. This will give more headroom for AIO-SX subclouds with a single numa node. Closes-Bug: 1923399 Change-Id: I433548792504f783a44a80d5099d93c5bee15ed7 Signed-off-by: Gerry Kopec commit 5b73ac5813e889ddcd0feb65f96eff0ac789e69a Author: albailey Date: Mon Apr 12 08:49:08 2021 -0500 Eliminate sdist step from sysinv zuul Zuul fails on setting up pbr randomly in the sdist step. It is unclear if the reason is that something is source packaged during tox which conflicts with another zuul job of a different interpreter, so simply disabling source dist generation. Most openstack projects are configured like this. Partial-Bug: 1922590 Signed-off-by: albailey Change-Id: Ieb4cf4113c07a0e166c001f124d920ba7118afb3 commit cbb9121a289603ec003dec098b8fa5918ca98300 Author: Marcus Secato Date: Thu Apr 1 16:30:13 2021 -0400 Adjust lock acquiring logic When locking the file descriptor skip_udev_partition_probe was not handling errors thrown by fcntl.flock which was leading controller-0 to degraded state after unlock. This change aims to strengthen that logic by handling the error properly, retrying the lock operation and improving logs. Closes-Bug: 1922256 Signed-off-by: Marcus Secato Change-Id: I000367668744a4e92e20ff9d3f1f8cd717883a46 commit 0e1bf356139abce842e61ede936d3f9a8ddc122e Author: Andrei Grosu Date: Tue Feb 2 20:55:27 2021 +0200 Send application lifecycle notifications for backup and restore. Implements backup, etcd-backup and restore hooks. Operations can fail so there is a second parameter, 'success' that is used to notify applications if an operation failed. Restore hooks are in place but not used by ansible playbooks. Separate semantic check action. Revert backup operations by keeping a list of all pre- operations with their associated 'revert' action and also, keeping in mind the logical order of the pre-backup and pre-etcd backup operations. Story: 2007960 Task: 40769 Signed-off-by: Andrei Grosu Change-Id: I0ebab45f4846cbcd25fecac6bf99195d9047eb8a commit d7909c05221ecf658fd92d63f0dfcfbe2a170e3b Author: Cole Walker Date: Tue Mar 30 14:09:25 2021 -0400 Function to create gnp after oam interface-network add Adds the initialize_oam_config function to manager.py which allows the correct flag to be set when creating a new oam interface. This flag is used to trigger the platform::firewall::runtime puppet manifest that sets up the hostendpoints and globalnetworkpolicy in kubernetes. Tested on: AIO-SX AIO-DX Standard Closes-Bug: 1911213 Signed-off-by: Cole Walker Change-Id: I34331ae6ad54ee4ce564467616c84931d2ae245a commit 1e97fb2398f1ecda43bfd4efb182647b4d719e90 Author: Mihnea Saracin Date: Fri Apr 9 18:27:40 2021 +0300 Add code to handle Kubernetes validating webhooks This commit will add new functions to list and delete 'ValidatingWebhookConfiguration' objects using the Kubernetes api. Partial-Bug: 1923185 Change-Id: I648e940f8104307e111213afd511f8fca19e39ab Signed-off-by: Mihnea Saracin commit f84ae1ecefb1e9e8f3f05216e80b6e0cdc36ef0c Author: Adriano Oliveira Date: Fri Mar 26 18:38:30 2021 -0400 sysinv-api OCF script API request In order to avoid other services requesting to sysinv-api before it is ready to handle request'; and to avoid individual services to implement their own retry logic, sysinv-api OCF script was changed to verify if sysinv-api is providing service before returning it is ready. Testing: 1. Bootstrap of AIO-DX 2. host-swact on AIO-DX 3. Upgrade path on AIO-DX, 20.06 to 21.05 (including host-swact) 4. subcloud bootstrap 5. Double sysinv-api kill causing swact In all tests, the logs are verified to confirm the retry logic is engaged and sysinv-api is properly started, also cert-mon. Closes-Bug: 1913455 Signed-off-by: Adriano Oliveira Change-Id: Ia17e9f7a15602c0cc52cb01896fac42ce4fcdcb9 commit 4314755481607a4aa7ea4ca25ccc5c22c2730956 Author: Andre Fernando Zanella Kantek Date: Tue Mar 30 13:40:43 2021 -0400 In AIO-SX, allow SRIOV VF parameters modify without host lock During runtime, user will be able to: modify an interface to be an SRIOV PF interface while specifying VF parameter --vf-driver and add new SRIOV VF interfaces while specifying VF parameters --vf-driver and --max-tx-rate Will not be permitted on an unlocked AIO-SX host to modify the VF driver and/or VF rate limit of an existing SR-IOV interface Story: 2008531 Task: 42203 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/784761 Change-Id: Icdb75df45ab2d6b31ecd60ab6c18e8042a7ee99b Signed-off-by: Andre Fernando Zanella Kantek commit 38c0ae47c3302ce956c65478281fccd1d8a50ebb Author: Robert Church Date: Thu Apr 8 00:59:50 2021 -0400 Deactivate app plugins when update fails prior to recovery When an application update fails and the previous version is recovered, the plugin infrastructure for the new app version is not fully cleaned up during recovery. If another application update is attempted, plugin activation/deactivation becomes out of sync and results in a KeyError when attempting to enable the new app plugins for update. This update will: - ensure that the failed app plugins are disabled prior to cleanup during application recovery - catch the KeyError if a similar situation occurs, log a message, and continue to cleanup the working set for the plugins Change-Id: If58942cd9342802bfd2055152c2f2d6289054084 Closes-Bug: #1923004 Signed-off-by: Robert Church commit 1c69d99d4ffb7d26296f0f3ef66830b4e64ee430 Author: Gustavo Santos Date: Fri Mar 26 18:11:47 2021 -0300 Restart tiller on openstack pending install check When the armada-api pod, which runs helmv2-cli, goes up, it connects tiller to a postgres instance running on the active controller. It does that using the active controller's floating IP address. On setups with more than one controller, this creates an issue where that connection is no longer valid after performing a host swact, since it still points to the old controller. After the swact, the first helmv2-cli command will fail with a broken pipe error. One of the steps in the Openstack installation involves checking the system for pending helm installations, which will run a helmv2-cli command, causing the application operation to fail if ran right after a host swact. This patch is a workaround to that problem. It forces a tiller restart by catching the first HelmTillerFailure exception caused by the broken pipe error and retrying the operation, which then will reestablish the connection between tiller and the correct instance of postgres. Closes-Bug: #1917308 Change-Id: Ia09f9d2844611471314d5d3af70e9bbb0938437c Signed-off-by: Gustavo Santos commit 3ad0dc17e7e5cd2de62f736db100a3fe6a38d928 Author: Yuxing Jiang Date: Fri Apr 2 14:05:57 2021 -0400 Accept additional attributes in addrpool-add CLI This commit adds additional attributes to enhance the current addrpool-add CLI. With this change, the addrpool-add CLI can create the addrpool with additional values including: 'floating_address', 'controller0_address', 'controller1_address' and 'gateway_address'. Test: Fresh install on an AIOSX, add addrpool with or without these attributes successfully by system addrpool command. Story: 2008774 Task: 42206 Signed-off-by: Yuxing Jiang Change-Id: I80955180e90accdcb6ce0c65dded2986f4a9c8ec commit 1128b30ce6abd878c1a3524fc584ddc42ba4cec4 Author: Isac Souza Date: Thu Apr 1 12:20:35 2021 -0300 New method to calculate replicas for armada apps Introduces _num_replicas_for_platform_app as a replacement for _num_provisioned_controllers which is used to set the number of replicas used by an platform armada application. The new method will use the same unrderlying logic but will never return a value less than 1. This will prevent having the replicas set to 0 when there are no provisioned controllers. Tested with unit tests and by checking the replica count after a host lock/unlock cycle. Partial-Bug: 1922278 Signed-off-by: Isac Souza Change-Id: If322ff5d02996c9b853bc350244899c5e22431a2 commit ead91ced1d55c986ec0f828124bfe7b7beb5ad59 Author: Melissa Wang Date: Wed Mar 31 12:13:43 2021 -0400 AIO-SX to DX: Check cluster-host interface config This update add a semantic check to ensure that the cluster-host interface exists and is not configured on loopback before allowing the AIO-SX to DX migration to begin. Story: 2008587 Task: 42188 Change-Id: Ic86ef966536ec41e495fb47e8e83682c1d03402b Signed-off-by: Melissa Wang commit bb6e8db56979f4ff741d6d7fe3304d1f9c45a99c Author: Robert Church Date: Tue Oct 27 17:12:56 2020 -0400 Add sysinv support for kubernetes to ignore isolcpus Based on the kube-ignore-isol-cpus host label which will generate /etc/kubernetes/ignore_isolcpus, also adjust the k8s_all_reserved_cpuset passed to puppet if the label exists. Basically if the label is set, we want to pass the isolated CPUs to kubernetes as though they were regular application CPUs. Story: 2008760 Task: 42167 Change-Id: I065ca40fbd3395bf86a02a7822c1f9d46ee3fe06 Signed-off-by: Robert Church commit d7c5a54ab94bce6635b83d91a807d28f97836a81 Author: Charles Short Date: Wed Mar 17 13:26:38 2021 -0400 sysinv: Cleanup requirements This commit does several things at once: - Drop argparse, its no longer required. - Add missing oslo.log, its been pulled in by other requirements but explicitly include it. - Use rfc3986 instead of Django, the requirement seem to be a bit heavy to me, to check for a valid URL. Added unittest. - Move run time requirements out of test-requirements.txt that are actually run time requirements for sysinv. - Also remove python2-mox from the centos7 spec file since its no longer required. - Updated debian/control and opensuse rpm as well. tests: - Did basic build test. - Ran smoke test with resulting RPMS and iso. Story: 2006796 Task: 42072 Signed-off-by: Charles Short Change-Id: I591ed52937092233c3ac7c2c2bf847b2b0c3f690 commit 9ed6740aebad0a6344e2490e89ade7afc3b26d9a Author: Charles Short Date: Tue Mar 23 11:49:05 2021 -0400 py3: Fix for python2/python3 compatibility - Replace 'range' with six.moves.range. - Replace 'zip' with six.moves.zip. - Reaplace 'map' with six.moves.map. - Replace a/b with a//b to use interger division on python3. - Replace dict.keys() with list(dict.keys()) to get a list on Python 3. On Python 3, dict.keys() now returns a view. test: - Changes were tested by running tox py27 and py36. - Build iso with added changes and ran some basic functionality Story: 2006796 Task: 41887 Signed-off-by: Charles Short Change-Id: I5086832605752bdb00a40a24596494c8fd987692 commit 4cc0fe7332af99a68d4a94115e3cb9ef8126a518 Author: Chris Friesen Date: Sat Oct 24 17:23:00 2020 -0600 add sysinv support for specifying cpu function by range In order to add flexibility we want to allow specifying CPU function by range, rather than just by count. This will allow us to run something like this on the CLI: system host-cpu-modify -f application-isolated -c 3-5,25 controller-0 There are a couple complications to be aware of. First, sysinv will NOT automatically add any missing SMT hyperthreads if the host has hyperthreading enabled. Second, when specifying CPU function for a different function via the CLI the range specification is lost and gets converted to a simple count. This implies that (for the CLI) only one function can support a range-based specification, and it must be specified last. Story: 2008760 Task: 42180 Change-Id: Id21d9968b6b0b59e163f42098be7a6f0e6ef739d Signed-off-by: Chris Friesen commit 012f5b82eee70fc838eba43729951491f8603f4e Author: MCamp859 Date: Fri Jan 22 17:05:07 2021 -0500 Remove SNMP APIs from user doc Deleted APIs for SNMP Communities and Trap Destinations. Added pointers to Fault Management guide. Story: 2008132 Task: 41396 Change-Id: Ic2c52bf1b11d1793d57c78264e757795af1deff3 Signed-off-by: MCamp859 commit 5a9ea65db6e5749de61100713994b937e2010438 Author: Dan Voiculeasa Date: Thu Mar 11 15:13:56 2021 +0200 Detect active/standby switch A swact can occur outside of a controlled maintenance action. An unexpected swact needs to be detected to let apps reapply evaluation happen. A swact is defined as follows: - sysinv-conductor runs only on active controller - hostname is used to determine where sysinv runs - sysinv-conductor switched controllers The detection can be run only once when conductor is restarted. The detection picks up both expected(system host-swact) and unexpected(reboot) swacts. A new trigger(APP_EVALUATE_REAPPLY_TYPE_DETECTED_SWACT) for evaluating apps reapply is used. Story: 2007960 Task: 42053 Signed-off-by: Dan Voiculeasa Change-Id: I48126887b0003a6a1a05ddd23f0429ef79a39832 commit 7343c40f99f861c30797b53b57c5afad58ac057d Author: Melissa Wang Date: Tue Feb 9 08:56:12 2021 -0500 Add support for AIO-SX to DX migration on subcloud This update allows the user to change the system_mode on a subcloud from simplex to duplex using the system modify command. The sysinv DB and the platform.conf will be updated with the new system mode. The semantic checks were modified to ensure that changing from duplex to simplex is prohibited. The changes also include support for updating the OAM networking config using the oam-modify command. Story: 2008587 Task: 41885 Signed-off-by: Melissa Wang Change-Id: If7c14222ca66323225400ed88f214655f33fe615 commit 3d42b28c77e47c4ce217aa746c786503df9d0818 Author: Andrei Grosu Date: Tue Mar 2 01:09:51 2021 +0200 Make the placeholder db entry unique for remote URLs. The original solution removed the 'app-name-placeholder' dummy entry if download failed but it was not approved because sysadmins expect this and manually removing the dummy placeholder app is advised. This patch simply sets an unique name for the placeholder using the first 16 characters for the URL md5sum as a postfix in case there are multiple application-upload performed in sequence ( by a script, for example) Closes-Bug: 1917374 Signed-off-by: Andrei Grosu Change-Id: Ib5db12bb23a0e7cce52596532e661d12092ea1d1 commit f8f826f960ca0359d0dfb62d7237beaf3ea07bcb Author: Andre Fernando Zanella Kantek Date: Thu Mar 11 05:52:11 2021 -0500 Allow modification of OAM IP addr, in AIO-SX, without locking the host Users will be able to change the OAM IP address without a lock/unlock cycle. To achieve this some services will be restarted (sm, sm-api, haproxy and vim-webserver) to reopen the L4 ports using the old IP address as part of the socket. Some config files in /etc are being updated also with the new address. Story: 2008531 Task: 42060 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/780955 Change-Id: I9e77fc60882f20d4f31c3e38b5305b1f207f40d9 Signed-off-by: Andre Fernando Zanella Kantek commit c11bd71fb1c7aadf9542cdd2a79d3c045c5c2b44 Author: Carmen Rata Date: Thu Mar 25 23:37:17 2021 -0400 Update psp migration script used in upgrade This commit updates the migration script used to apply the pod security policy (psp) configuration during an upgrade. The change is necessary to accommodate newly added psp ClusterRoleBinding configurations. Closes-Bug: 1885716 Depends-On: https://review.opendev.org/c/starlingx/ansible-playbooks/+/782325 Change-Id: Ibfdfd51e588eb2ad47c9f1c116875d01a2f06502 Signed-off-by: Carmen Rata commit 3f6c732939b8fc857c932ee05b3418545ccae8f1 Author: Andrei Grosu Date: Fri Feb 12 16:20:37 2021 +0200 Regenerate the correct plugins path. For some reason, if the application to be uploaded is a (remote) URL, parsing the manifest if deferred (presumably not to block for large files and/or slow networks) and a dummy 'app-name-placeholder' is used until later, when the file is unpacked and its manifest read. Closes-Bug: 1915518 Signed-off-by: Andrei Grosu Change-Id: Ic3929965ea931b117c3e5aab6f8e3f128bbbeb56 commit e2e7d45d0e5a84f7ef345a5a6561c5370a85a5fd Author: Tao Liu Date: Wed Mar 24 20:35:22 2021 -0400 Fix dcorch subcloud audit issue after the upgrade Dcorch-engine stops auditing the subclouds after the upgrade. This is because audit_status of subcloud_sync data was not set during data migration. This update sets audit_status to initial state of “none”. Test: Upgrade controller-1, and then swact to controller-1. Verify the dcorch-engine audits subclouds. Closes-Bug: 1920962 Signed-off-by: Tao Liu Change-Id: If8fa6c5e1c1d1a81104976cb3e527c4095dd97f7 commit ed127df6ad9b8863080ef6cf5f44be2e5966518f Author: Charles Short Date: Mon Mar 8 09:01:03 2021 -0500 Remove oslo-incubated version A long time ago oslo-incubated code was used to build the individual projects the same way. Now the openstack projects use pbr to build the python projects. Remove oslo-incubated version code, it is not being used anywhere so just remove it. Unit tests run fine when this module has been removed. Story: 2006796 Task: 42010 Signed-off-by: Charles Short Change-Id: Ib11d69210878f38febf2d031b083a1ad85fec30c commit 08c14894f3bf7fb6645707f689394dded27800ea Author: Teresa Ho Date: Wed Mar 10 09:45:24 2021 -0500 Add bond option primary_reselect This update is to allow the option primary_reselect configurable for aggregated ethernet interface. The option is to prevent reverting between the primary slave and other slaves. Story: 2008706 Task: 42057 Change-Id: Icacc0bd2d5e42bf2e5db1505fd676c628dbe3ed1 Signed-off-by: Teresa Ho commit 7dcdfaae898960c8f6ea7dbee892c118a1b80318 Author: Andrei Grosu Date: Mon Mar 1 20:45:06 2021 +0000 Progress adjust metadata refactoring Changed the name of the constant and the yaml key to better reflect the purpose. Now the value is an integer which represents the adjustment value used to compute the percentage completion when applying charts. Cleaned up the code around the usage of the value and computing the percentage. Story: 2007960 Task: 41959 Signed-off-by: Andrei Grosu Change-Id: Ia3b07b83762cdf20f6809222dc687f67c15deee5 commit 368f5ce3217314acea4253cf2ade25d9d6684580 Author: Dan Voiculeasa Date: Thu Feb 11 12:51:31 2021 +0200 Enhance maintenance semantic checks with app hooks Let apps run semantic checks for lock and unlock actions. Let forced actions not run the app semantic check. Create unit tests for allowing and rejecting the action by an app. Story: 2007960 Task: 41842 Signed-off-by: Dan Voiculeasa Change-Id: Ibe35c917cd5702031a56baf3059b70e0e2e59480 commit b3b28f56e8b52727575667b62a9d10dd409d7d18 Author: Suvro Ghosh Date: Fri Mar 5 10:40:14 2021 -0500 Force remove or delete application Adding the functionality of using the flag -f or --force with system application-remove or system application-delete Story: 2007960 Task: 42016 Signed-off-by: Suvro Ghosh Depends-On: Ia1017b7eff0d9bd73b6448f2c4790f7e2b89c828 Change-Id: If68d66d799addcd996da4b146d092c855b455aa3 commit 8f0312b5184f9292dbea878d8b4da2e3ef2e786c Author: Daniel Pinto Barros Date: Fri Feb 19 10:31:43 2021 -0500 Introducing GEO location new fields for System New fields was created for the system object. Changes was made to include GEO location attributes (latitude, longitude) to the system object and adding a way to retrieve and modify those attributes using the API and CLI. Updates on: DB system model; DB migration; System object fields; API fields; CLI fields; API documentation. Story: 2008570 Task: 41721 Signed-off-by: Daniel Pinto Barros Change-Id: I86f124c44d80896427e3ac1bc799fe34588ae942 commit fa1622ef5f80950928f8f2447f9c0e178e797704 Author: Suvro Ghosh Date: Fri Mar 5 13:23:14 2021 -0500 Prevents critical apps from being removed If an app has a metadata stating remove is prevented/forbidden then "system application-remove" for that app will be rejected Story: 2007960 Task: 42005 Signed-off-by: Suvro Ghosh Change-Id: Ia1017b7eff0d9bd73b6448f2c4790f7e2b89c828 commit da2bc95fbcc359f9ad4e719573f19fc2bcd0afd1 Author: Mihnea Saracin Date: Thu Mar 11 13:28:55 2021 +0200 Change enable_secured_etcd.yml variable in upgrade The enable_secured_etcd.yml playbook will use the cluster_floating_address variable instead of default_cluster_host_start_address. So we change the upgrade script accordingly to use the new variable. Closes-Bug: 1918130 Depends-On: I8fecc1e5e54b5a9a9a72a54c069f79f5f2d434ba Change-Id: I8c9fd36e1104d4713bb748a57193530a0c4b458a Signed-off-by: Mihnea Saracin commit cc27551a8e797398527a18815838ff467c7c0a39 Author: Charles Short Date: Wed Feb 17 13:05:30 2021 -0500 Remove unsafe umask usage The sysinv code runs under eventlet that causes the running greenthread to swap out the original umask. This results in the sysinv code running with the incorrect umask. This can be demonstrated by the "system dns-modify" command, the agent process starts with a umask of 022, switched to 0, and is never restored. This simple fix is to audit where os.umask is being used and replace os.umask with os.chmod. Testing has been done locally by: 1. Running the "system dns-modify nameservers=8.8.8.8,8.8.4.4" command and checking the results/permissions of /etc/resolv.conf. Also, cheked to see if the umask flag have been changed in /proc/XXX/status before and after running the command. 2. Running an "system applicaton-upload" command on an installed helm armada package, these are located in /usr/local/share/application/helm. After the application upload, the application-apply should be "applied" without error/failure as shown in "system application-list". 3. Running a distributed-cloud and checking for any errors. The command "dcmanager subcloud show " should show the identity service in sync after the dcmanager subcloud manage <>". Closes-Bug: 1915955 Signed-off-by: Charles Short Change-Id: I16ce695cfc4f6fb496ac0b3287906cc968ec5e98 commit c72417aedeb9d300ca256289a596a5797c863b51 Author: Dan Voiculeasa Date: Mon Feb 8 13:27:35 2021 +0200 Implement algorithm for reapply evaluation priority Implement algorithm to determine app priorities for reapply evaluation. Use information provided in metadata to create a directional graph. Detect cycles and abort. Unit tests added. Tests: AIO-SX, AIO-DX Apps are correctly ordered for reapply evaluation. Applications reapply order: [u'cert-manager', 'rook-ceph-apps', 'platform-integ-apps', 'oidc-auth-apps', u'stx-openstack'] Story: 2007960 Task: 41781 Signed-off-by: Dan Voiculeasa Change-Id: I375a90b746a0ff4c970305a26c2e3e061b14454e commit 4face8a656ac63d110a2c1d8ce4efd98af771c47 Author: Dan Voiculeasa Date: Tue Mar 2 22:03:53 2021 +0200 Rework platform managed apps At the moment the managed apps are hardcoded. This behavior needs to be changed. Let apps specify in the metadata if they want to be managed or not. Let apps specify in the metadata the state they want to achieve. Create column in kube_app table to store metadata. This will be read when conductor is restarted. Tests: Install AIO-SX and AIO-DX, apps achieve the state described in their metadata file. Restart conductor, metadata gets picked up from the database. Do system application-remove, app gets auto-applied. Do system application-delete, app gets auto-uploaded. Story: 2007960 Task: 41780 Signed-off-by: Dan Voiculeasa Change-Id: I881716049471183cfd1179ab0558a557c8d104d8 commit a87694bf5e0362a5466b84c4c6638ffd453e9520 Author: Teresa Ho Date: Thu Mar 4 11:10:31 2021 -0500 Use http_port from conf file for fpga-agent The sysinv-fpga-agent is modified to use the http_port parameter from the platform.conf file. If a device image update operation is in progress, the http_port service parameter modification cannot be applied. Story: 2007875 Task: 41969 Change-Id: I41e795606535d91131b96a014b07bf18f0032d57 Signed-off-by: Teresa Ho commit ff3ce494eea4fb37a97897a7cceac70e60b2f727 Author: Chris Friesen Date: Fri Feb 26 17:01:26 2021 -0600 Notify dcmanager when upgrade completed When an upgrade has been completed we want to notify dcmanager so that it can do a load audit of the subclouds rather than waiting up to an hour for the normal load audit to run. Story: 2007267 Task: 41967 Depends-On: https://review.opendev.org/c/starlingx/distcloud/+/778338 Change-Id: I0c03bbfa16745fa297e159256a284e8862ff926a Signed-off-by: Chris Friesen commit b9cd8ec6de2b86a966c062c0db45a8b90c041526 Author: David Sullivan Date: Thu Mar 4 09:11:30 2021 -0600 Upgrade activation interrupted by host-swact During an upgrade-activate the upgrade scripts can be interrupted by a swact. We need to block the swact during the activation. If a swact does occur we need to reset the upgrade state so the activate can be attempted again. Closes-Bug: 1917779 Change-Id: I9274319375296b2334533e386629d185e2b472ac Signed-off-by: David Sullivan commit 69b62362c75cba883446499c5a97a984293d2eca Author: Litao Gao Date: Thu Mar 4 03:19:06 2021 -0500 Update api-ref with modified interface configuration 1. allow creation of ethernet interface using sriov interface 2. max_tx_rate options for sriov vf interface configuration Story: 2008470 Task: 41987 Signed-off-by: Litao Gao Change-Id: Id64248060b57d1778a455637ba9cf70d680456e5 commit 505e93b4c0d34ddf9e3d24769300d9adeace8da7 Author: Isac Souza Date: Wed Feb 17 20:38:44 2021 -0300 Fix handling of expired watch event in cert-mon. The old code did not account for a type='ERROR' being received by the watch stream. The new code checks if the received event is an error and returns from the infinite loop to start the watch from scratch. Closes-Bug: 1914408 Signed-off-by: Isac Souza Change-Id: I7feabe5b550979d3761427ae501f1a94903a8983 commit 38cc2e8f4f266125fb08a71e3531a774b433f485 Author: Tao Liu Date: Tue Mar 2 16:17:18 2021 -0500 Allow routes to be configured on oam interfaces This update adds NETWORK_TYPE_OAM to ALLOWED_NETWORK_TYPES for configuring routes. Test: Configured routes(add/delete) through the system CLI against the OAM interface. Story: 2007267 Task: 41971 Signed-off-by: Tao Liu Change-Id: I053ce52a760fb2b20d81d0b250bde6d9902ddaa2 commit 240be6fee8cedec4a0f36d1d624716b0766bcea2 Author: Chris Friesen Date: Mon Mar 1 17:50:38 2021 -0600 move rest_api to common code The rest_api.py function is potentially useful for all sysinv code, so move it to the "common" subdirectory to make this more clear. Story: 2007267 Task: 41967 Change-Id: I3ef2b4144f85ad6e3533e0236f4afb83bdae707e Signed-off-by: Chris Friesen commit 890b1208ca2ff67ad32228be4be7671f673f4a90 Author: Babak Sarashki Date: Wed Feb 17 15:28:19 2021 +0000 config: Add global service parameter to set cri handler This commit adds global service parameter "container_runtime" to allow setting container runtime interface (CRI) entries in the containerd configuration file for custom runTimeClass. An example usage to set the cri: system service-parameter-add \ platform container_runtime \ custom_container_runtime=my_crihandler:/absolute/path/to/my_criBinary Story: 2008434 Task: 41390 Signed-off-by: Babak Sarashki Change-Id: Icc5fd16682f4cf47abff16e20a5332fc195c4afc commit 68bb5ecc2b1de4ad1ece4280f538d9ad01442bb8 Author: Litao Gao Date: Fri Feb 12 11:45:15 2021 -0500 Fix miscalculation of the available link speed bandwidth Previous implementation has two issues: 1. forgot to multiply the max_tx_rate by vf number 2. missed the case of vf subinterface modification Story: 2008470 Task: 41508 Signed-off-by: Litao Gao Change-Id: I4d62ed177124fb2c4ddb547a37d9ea988a332dbd commit 790b758fe4a87227cd6b76a97b301df744f6b53f Author: Litao Gao Date: Sat Feb 13 02:40:15 2021 -0500 Need to check MTU setting for the ethernet type subinterface Previous implementation only need to check the MTU for VLAN and VF type subinterface, since we've introduced ethernet type subinterface which can be created on pci-sriov interface, also need to add the MTU check for ethernet type subinterface. Story: 2008470 Task: 41505 Signed-off-by: Litao Gao Change-Id: I1e38cb63496013539b91c749076e4cbb5a951bfc commit e86cc068932f953b34ba212e21dc7849f0f75325 Author: Litao Gao Date: Tue Feb 9 10:59:45 2021 -0500 Remove the checks which restricts vf subinterface creation/modification Allow pci-sriov class and vf type subinterface creation/modification using the pci-sriov interface when it has already been used by other VLAN interface Story: 2008470 Task: 41505 Signed-off-by: Litao Gao Change-Id: I51d936530a93e3a9e6ac81586c16df4b7342a180 commit b27f224803741b7e9b236671643acbde238578ac Author: Suvro Ghosh Date: Thu Feb 11 13:40:47 2021 -0500 Prevents critical apps from being deleted. If an app has a metadata stating deletion is prevented then system application-delete will be rejected Story: 2007960 Task: 41882 Signed-off-by: Suvro Ghosh Change-Id: I4401d3af6e7af354783edc945c1a5cdb72c1d0a1 commit 0304082c8918470d536064157f9cad5902823a1b Author: Dan Voiculeasa Date: Mon Mar 1 19:12:17 2021 +0200 Fix logging during reapply evaluation If an app is in applied-failed or applying states the evaluation of reapplies the flow of execution is stopped. Not all applications might get the evaluation done. Introduced by I023eb3bce9061e0ccfcf10ebeeaef91bcb39cff1. Story: 2007960 Task: 41760 Closes-Bug: 1884770 Signed-off-by: Dan Voiculeasa Change-Id: I604beb78cc7112c9d05bec358787b74aa28ebe14 commit ea1cec6cd5812a383153b07ba6b31a219118fa8b Author: Teresa Ho Date: Fri Feb 26 14:58:42 2021 -0500 Upgrade script to create device_images directory Added an upgrade migration script to handle the case where the N-side does not have the /opt/platform/device_images directory. Upgrade testing was performed. Story: 2007875 Task: 41942 Change-Id: I42bd944b831243ddfc35a76309be095008ec749d Signed-off-by: Teresa Ho commit a86f69a5b401f81b91fb4f9feb4ce95ee03510ff Author: Matt Peters Date: Mon Mar 1 09:12:37 2021 -0600 Config API documentation for Kubernetes cluster API Reference documentation for the Kubernetes Cluster API for cluster access information. Story: 2008630 Task: 41914 Signed-off-by: Matt Peters Change-Id: Id0942dafeb2273e271d145f20517d16b1f409560 commit 6add4f2dfbb162dc21815489bf5f24b0b9eb6e7f Author: David Sullivan Date: Fri Feb 19 09:21:32 2021 -0600 Support background runtime manifests during upgrade-activate In distributed cloud environments runtime manifests can be applied in the background. This can cause hosts to become config out-of-date after the upgrade-activate completes. This is due to the large window between setting the host's config_target and updating the config_applied. If a manifest is run in this window the host will remain config out-of-date until a lock/unlock is performed. To address this the config_target changes will be limited to hosts that apply a runtime manifest as part of the upgrade-activate process. Further the config_target will be updated immediately before the _config_apply_runtime_manifest is called. Story: 2008055 Task: 41917 Change-Id: I2e60c7557e8d398eeef2a407a0552f5e8f4a1f18 Signed-off-by: David Sullivan commit b08ad875ebc9f1c6383b1ad48e9bee19b7425f94 Author: Matt Peters Date: Tue Feb 23 08:33:18 2021 -0600 Config API for Kubernetes cluster access information Introduces a new sysinv config API for retrieving the Kubernetes cluster access information and security credentials (if configured). The information may be used to configure a remote Kubernetes client with the required configuration to use the Kubernetes API with administrative privileges with either client certificate authentication or token authentication for the kubernetes-admin service account. The following information is available for each cluster: Kubernetes Cluster Name (kubernetes) Kubernetes Release Version Cluster API Endpoint URL Cluster Root CA Certificate Admin Client Certificate Admin Client Key Admin User Name (kubernetes-admin) Admin service account token Story: 2008630 Task: 41836 Signed-off-by: Matt Peters Change-Id: Ib81c7fcc3a577c1209ab3a0dd882552ba3d2b9db commit c4affcaf43ff936f974b3650097bbd6beac93a0c Author: Litao Gao Date: Tue Feb 9 04:36:12 2021 -0500 Remove the checks to enable flexible pci-sriov interface config ordering 1. Allow pci-sriov class interface creation without mgmt. interface configured 2. Allow interface transition from platform to pci-sriov class Story: 2008470 Task: 41505 Signed-off-by: Litao Gao Change-Id: Icd22ed5ae8314e8849dc07e3fc94baac26f49079 commit df6a3386544a2c1defcf4a9cc35f3eda0daa1e4f Author: John Kung Date: Mon Feb 15 15:05:24 2021 -0600 Ensure agent is ready before issuing runtime config Update the conductor to ensure at least its own config agent on the active controller is ready to handle the config. Otherwise, append runtime config to deferred config list until signalled ready. Prior to this commit, runtime config can be missed on startup on the active controller due to the agent not being ready to handle the config request. It does not defer runtime config application until all other hosts are ready as the config target is still persisted to track required config for host target in the event the rpc request is missed due to unexpected event or agent not ready on other hosts. Test Performed: Installation and deployment of AIO-SX Installation and deployment of multinode system Verify that deferred runtime configs are applied in order when agent becomes available. Change-Id: I7388844d048453d302409eea36a939d81c9447ec Closes-Bug: 1915343 Signed-off-by: John Kung commit 96a51f34fafcc71fba2cbec25067a822120c4a6c Author: Teresa Ho Date: Wed Feb 17 16:32:40 2021 -0500 Retrieve device image over lighttpd Modify the fpga-agent to fetch the device image from the controller via http instead of the drbd directory. Tests performed on the following systems: AIO-DX, AIO-DX plus compute, Standard 2+1 DC with AIO-DX plus subcloud DC with Standard subcloud Upgrades on duplex system Story: 2007875 Task: 41879 Depends-on: https://review.opendev.org/c/starlingx/stx-puppet/+/776490 Change-Id: I9a53eb2131c5ce2c2b87c1740e234af65ffabf78 Signed-off-by: Teresa Ho commit 61dfb4c0e61fc7aa3a8c33e6a5984525b4c9d608 Author: Andy Ning Date: Fri Feb 19 13:40:12 2021 -0500 Enhance upgrade script 85 to handle empty data Upgrade script 85-update-sc-admin-endpoint-cert.py extracts admin endpoint certificate and private key from cert-manager. But intermittently the certificate data returned is empty, causing an empty /etc/ssl/private/admin-ep-cert.pem generated and outage of many services. Eventually the upgrade is failed. This update enhanced the script to detect invalid data returned and retry to extract the data from cert-manager. Change-Id: Ib2c7da9147f28bf10dcb8a053412a8f94af42353 Closes-Bug: 1916279 Signed-off-by: Andy Ning commit 5cc9dc9c86b9214803ebba1231e227af862db77a Author: Dan Voiculeasa Date: Fri Jan 29 11:10:43 2021 +0200 Rework evaluating apps reapply At the moment reapply evaluation is performed only for hardcoded apps. This behavior needs to be changed. Allow apps to specify what triggers they subscribe to in their metadata file. In sysinv a trigger is a dictionary composed of at least the 'type' field and other optional fields. Easy use of filters on the triggers by specifying key:value pairs to be searched in the trigger information. Keep a map between sysinv triggers type and the trigger types specified in the metadata. Existing triggers are: host-delete, runtime-apply-puppet, platform-update, unlock, force-unlock. Add new triggers: host-add, host-delete, host-reinstall, lock, swact, force-lock, force-swact, system-modify. Introduced a lifeycle semantic check hook for reapply evaluation where apps can run complex logic to reject the evaluation. Modified unit tests. These are basic unit tests querying the number of calls. Unsupported unit test for host-add. Missing unit test for system-modify. Tests: AIO-SX, AIO-DX Triggering the events, events are seen in the logs as expected, apps respond to events as specified in their metadata. To keep backwards compatibility, this work depends on updating the metadata file for each of the 5 apps that use the existing triggers. Depends-On: Ia7bbca906e343ffffa019885a790befdf5ccb565 Depends-On: I588648090e82ac573db5112a2704d22fa45a049f Depends-On: Ie02743cdf056dda3feb66911c74f9dabe69d98dd Depends-On: I4778c6c8232fffb5fafca95b450e590fbb1b0f64 Depends-On: I0a76ef10fe3958634d714f1484d79763f98a0d4e Story: 2007960 Task: 41760 Closes-Bug: 1884770 Signed-off-by: Dan Voiculeasa Change-Id: I023eb3bce9061e0ccfcf10ebeeaef91bcb39cff1 commit fb5956ed3aa068a2b6cd7851dcefd2bdce0244ff Author: Angie Wang Date: Tue Feb 23 18:18:53 2021 -0500 Update the minimum small root disk size Dc-vault filesystem should be counted into the minimum root disk calculation. Partial-bug: 1916797 Depends-on: https://review.opendev.org/c/starlingx/metal/+/777464 Change-Id: I65ac2cc5bda3a94728a7f593b6aadbafca7a3af6 Signed-off-by: Angie Wang commit 2b7a745b7f033de13ca6fb42afcb6a8929afdc45 Author: Dan Voiculeasa Date: Fri Oct 9 15:07:16 2020 +0300 Migrate to database backend for backup and restore Drop the use of a flag file for restore. Create the database table for backup and restore. Use the database to store information. CREATE TABLE backup_restore ( created_at TIMESTAMP, updated_at TIMESTAMP, deleted_at TIMESTAMP, id serial PRIMARY KEY, uuid VARCHAR ( 36 ) UNIQUE NOT NULL, state VARCHAR ( 128 ) NOT NULL, capabilities TEXT ); This is an improvement to allow backup and restore functionality to be extended in an easier manner. Not required to fix the the bug. Added unit tests. Depends-On: I7b7fab99d457056032dbbd612363cd5036736cda Depends-On: I44fc4aaa528e372a84115714f271b4f5e063f86e Partial-Bug: 1887648 Change-Id: Ibb96696a35fe7b560aa002c442e2af735d08ec24 Signed-off-by: Dan Voiculeasa commit fca3000f725bb1ec447a7546c7d3805723544352 Author: Isac Souza Date: Wed Feb 17 22:06:21 2021 -0300 Fixes the time calculation in the is_expired method of the Token class The previous implementation calculated an absolute delta between the expiration time and the current time and compared it to a time window. The calculation should be done without getting the absolute value to account for an expiration time older than the time window. Added unit tests for the affected code. Closes-Bug: 1915952 Signed-off-by: Isac Souza Change-Id: I46bb36993d6f02978a593a1c1d49692e627d2a9b commit 00d26663ad0141277620d438fb5904d3de715077 Author: Charles Short Date: Tue Feb 16 11:17:32 2021 -0500 Deprecate sysinv.openstack.common.rootwrap Deprecate sysinv.openstack.common.rootwrap in favor or oslo_rootwrap. This was done so we maintain less code and we worry about Python3 a little less. Story: 2006796 Task: 41868 Signed-off-by: Charles Short Change-Id: I979ff1c8045030cfcaf8b88678121c4d0d684743 commit 063d9f84487407242ac1eb6cded2062abfdc01f0 Author: David Sullivan Date: Thu Feb 11 10:01:52 2021 -0600 Limit rootfs size constraints to controller nodes Previously rootfs size restrictions were limited to controller nodes. Return to this model to better support lab and virtual installs. Closes-Bug: 1915215 Change-Id: I6dd53f74de54c971c59c5d821ec33a33c7e152f5 Signed-off-by: David Sullivan commit 6ab90d747a953f5a9f75bc05d0b55d3e2ac74fd2 Author: Babak Sarashki Date: Tue Jan 5 16:08:43 2021 -0500 sysinv: Intel ACC100 (Mt Bryce) enablement This commit adds SR-IOV device plugin support for forward error correction (FEC) devices that are enabled on an Intel ACC100 (Mt. Bryce). The Intel ACC100 is mounted on Lisbon ACC100 Card. The FEC device is intended for use by a DPDK application. It is presented to the system under resource name: "intel_acc100_fec." An example usage to modify the device: system host-device-modify \ -e true \ --driver igb_uio \ --vf-driver \ -N And example assignment to a pod: resources: requests: memory: 4Gi intel.com/intel_acc100_fec: '16' cpu: 6 limits: hugepages-1Gi: 2Gi memory: 4Gi intel.com/intel_acc100_fec: '16' cpu: 6 Story: 2008440 Task: 41403 Depends-On: https://review.opendev.org/c/starlingx/stx-puppet/+/775253 Depends-On: https://review.opendev.org/c/starlingx/integ/+/775252 Signed-off-by: Babak Sarashki Change-Id: I831fd16a0410ee988365c067789f760139274ec8 commit a45569eda4daf7f323294a116a76d1bb75e1e5ed Author: Cole Walker Date: Tue Feb 16 20:54:11 2021 +0000 Revert "Add api call to trigger creation of gnp" This reverts commit cd344669d381103b8e065617811b928eb895eacb. Reason for revert: Config out of date alarm is present after install. Code rework is required. Change-Id: Ic71eeb8b9d95436b1d1999bf3a1d28022a972077 commit 0786d37bebf244c918dd227fe3112af80870cc79 Author: Andre Fernando Zanella Kantek Date: Wed Feb 10 08:43:48 2021 -0500 Allow SR-IOV interface modify on AIO-SX without locking the host On an unlocked AIO-SX host, it will be possible to modify a non-SR-IOV interface to SR-IOV PF, and will also be possible to add and delete SR-IOV VF interfaces But the user will not be able to modify parameters on an existing SR-IOV interface as well modify an SR-IOV PF to a different class Story: 2008531 Task: 41802 Change-Id: I1b8b214816404fb0893e77acd55798017fa836bc Signed-off-by: Andre Fernando Zanella Kantek commit 625017798a93d2fe57740c981fe3ba7c6acaebae Author: Bart Wensley Date: Wed Feb 10 15:06:03 2021 -0600 Add unit tests for _create_host_filesystems Adding unit tests for _create_host_filesystems in the sysinv AgentManager class. This points out the issue described in LP1915215 and will ensure any future changes don't break existing behaviour. Also clamping the version of astroid to allow pylint to pass properly in local test environments. Change-Id: Ifd4da0e044c3262281e52826854c20e0d161a3bc Partial-bug: 1915215 Signed-off-by: Bart Wensley commit 500d4e250c49c92b1b3440d068ab654483a437fc Author: Carmen Rata Date: Mon Feb 8 10:35:26 2021 -0500 Fix out-of-date alarm wrongful removal When updating docker service parameters by either adding or deleting them, a "Configuration is out-of-date" alarm will automatically appear for both active and standby nodes. The alarms should only be cleared by a manual lock/unlock of each node. The fix addresses the scenario where a configuration update that triggers a runtime manifest clears the existing out-of-date alarm (e.g.: system certificate-install -m ssl_ca ). The issue is reproduced no matter what configuration update with runtime manifest is used or if the existing alarm was due to a “service-parameter-add” or a “service-parameter-delete” command. Closes-Bug: 1878018 Signed-off-by: Carmen Rata Change-Id: I1e545327ae394385c995aee885590779432c60b0 commit f2908be215b6c86fb273587c34ec23360791d276 Author: Isac Souza Date: Fri Feb 5 22:25:45 2021 -0300 Backward compatibility changes for registry cert managed by cert-manager If a user tries to install a docker_registry certificate and that certificate is being managed by cert-manager (i.e. secret is present), we will refuse to install it unless a force=true parameter is passed in. Similar to the logic already present for mode=ssl. Story: 2007361 Task: 41782 Signed-off-by: Isac Souza Change-Id: Ief4ba51054e6fec951f4c44dbf40c2b3e2e8b292 commit 983add3417b2b5dbeadc22be1f1997ca9e0ab16e Author: Isac Souza Date: Thu Jan 28 12:44:09 2021 -0300 Docker registry certificate management by cert-manager Cert-mon changes to monitor 'system-docker-local-certificate' k8s secret and install StarlingX docker_registry certificate for registry.local:9001. Changes include: - New thread to watch registry certificate changes - Refactored the code to reduce code duplication - Call sysinv api for 'certificate_install' Design testing completed: - When k8s secret is added/modified (initiated by cert-manager), certificate installation is completed - sysinv api 'certificate_install' installs & confirmed via openssl s_client -connect registry.local:9001 - When certificate is renewed, keys get regenerated (no changes needed. Confirmed that existing infrastructure takes care of it) Story: 2007361 Task: 41717 Change-Id: Iffa68486764287a1b82a183ab9801a53c1e4885b Signed-off-by: Isac Souza commit c69304a998f6bc5b76b31c161ae4b6815ad60b54 Author: Gonzalo Gallardo Date: Thu Feb 4 16:29:26 2021 -0300 Restore system modify command for SNMPv3 Restores "system modify" command for --(contact|name|location) options. Also, this affects the "system_name" variable in [fm.conf] configuration file. Story: 2008132 Task: 41761 Signed-off-by: Gonzalo Gallardo Change-Id: I143873953603b36c5e79f5247801b33efe464022 commit 4cd77f035a812d8fbc6f060ca483b8050c65cb91 Author: John Kung Date: Mon Feb 1 15:23:04 2021 -0600 Update config tracking for reboot required config The reboot config tracking needs to be updated to ensure it aligns the target with the actual applied config. It must also initialize and only clear the config for the active controller if a reboot has actually occurred. On host-swact, the target controller is checked for config up to date condition, rather than both source and target controller; as the source active controller would still tracked the config requirement in the persisted database. Improve traceback of config requests. Tests Performed: Perform host-swact with active controller config out of date Perform host-swact with standby controller config out of date Verify AIO-SX and Duplex config operations Restart sysinv with reboot config set and perform runtime config Perform runtime config while reboot required config set Perform runtime config while reboot required config cleared Change-Id: I339ad82a2c7b37ac1c97c3eb790a231a40914250 Closes-Bug: 1914085 Signed-off-by: John Kung commit 36a4ff4fd2194bf876c1ce68c17f16b047bff7ae Author: Adriano Oliveira Date: Mon Feb 1 19:48:28 2021 -0500 Fix migration scripts execution sequence The following changes have been introduced as a fix for this issue: 1. Changed the sorting on the migration script file names to be based on the first number on the file name. 2. Added file name format validation: "nnn-*.*", where "nnn" string shall contain only digits. 3. Fixed the name of two migration scripts that were not following the correct format (not using "-" separator). 4. Added set of unit tests to test and validate the execution of migration scripts code. Manual upgrade testing to STX 5.0 has been executed. Closes-Bug: 1887985 Signed-off-by: Adriano Oliveira Change-Id: I04fdb8a3b3e177c609c4037825810a531954d99c commit d64f6f59244c56f26899aa5f2b8fb8a33714f3b7 Author: Andre Fernando Zanella Kantek Date: Wed Jan 27 13:43:59 2021 -0500 Support data network assignment on an unlocked host For SR-IOV interfaces, when operating in AIO-SX, it will be possible to assign it to a datanetwork without host lock Story: 2008531 Task: 41705 Change-Id: Ia4b5670ffe09255845823dc3ae2c3fc19c709fc9 Signed-off-by: Andre Fernando Zanella Kantek commit cd344669d381103b8e065617811b928eb895eacb Author: Cole Walker Date: Tue Jan 12 10:25:18 2021 -0500 Add api call to trigger creation of gnp Creating an oam network was not correctly setting the flag to trigger the runtime puppet manifest to create the default globalnetwork policy and hostendpoint resources. This change adds the correct api call to ensure that the flag is set when an oam network is created. This ensures that the default globalnetworkpolicy and hostendpoint kubernetes resources are created as intended. This issue was previously masked by the fact that the relevant runtime manifest was being run on every single unlock. This behaviour was changed and revealed that the required flag was not being set upon oam creation. Closes-Bug: https://bugs.launchpad.net/starlingx/+bug/1911213 Signed-off-by: Cole Walker Change-Id: I5b570e82246d82a4243ee68a871333876d231c85 commit a4a969c94f2ef4dc1531592b6563e90d6465d6ac Author: Mingyuan Qi Date: Tue Nov 24 02:13:09 2020 +0000 Add host command support for the edgeworker node This is an experimental feature in stx5.0. This commit enables following commands for the edgeworker node: system host-add/system host-update/system host-delete After the host being added/updated, the mgmt ip of an edgeworker node will be assigned during the configuration process of it. There will be limitations of edgeworker nodes before the final phase of the feature finished: - The Kubernetes provisioning requires ansible playbook triggered manually. - Gather node HW information is not supported. - Configure node from controller is not supported. - Manage node lifecycle is not supported. - Update/upgrade node is not supported. Story: 2008129 Task: 40862 Change-Id: I7e6de65ba848d9468a4e5afddd16b1cd9e3cd7dd Signed-off-by: Mingyuan Qi Depends-On: https://review.opendev.org/c/starlingx/config/+/761716 commit aa851db31fa3ea25337da819d6fd9fc874f7fcfd Author: Martin, Chen Date: Fri Apr 17 11:02:08 2020 +0800 Introduce rook ceph 1, add storage backend rook ceph for container-based ceph cluster 2, create puppet rook.py 3, update sysinv-agent audit function rook ceph provision disk with bluestore, which create a vg and lv update sysinv-agent audition function, so after rook-ceph applied add rook-ceph created vg named ceph-xxxxx to sysinv-db 4, update lvm filter for rook-ceph provisioned osd Story: 2005527 Task: 39452 Depends-On: https://review.opendev.org/#/c/713084/ Depends-On: https://review.opendev.org/c/starlingx/rook-ceph/+/716792 Change-Id: If8c1204dd3c7cc25487b2f645ace9aa680d32d59 Signed-off-by: Martin, Chen commit 8671cb8f5a8340bdd5f623343bde39f76e9dc3fc Author: John Kung Date: Thu Jan 21 09:14:18 2021 -0600 Upgrades: data migration for dcorch scaling The dcorch scaling feature introduced new tables, of which the subcloud_sync table needs to be populated. The subcloud_sync data is populated with initial values similar to what would have been done on subcloud add. Change-Id: Ib880ea19cb7f69cb004d58557ad07535c1e45774 Story: 2007267 Task: 41648 Signed-off-by: John Kung commit b40f98b57e3274892c893d30139e03bb6b21d6d2 Author: Litao Gao Date: Mon Jan 25 03:46:08 2021 -0500 Interface profile support for VF rate limiting configuration This feature is part of 'Single NIC support'. This commit eanbles interface profile creation and apply of interface configuration which includes VF max-tx-rate config. Story: 2008470 Task: 41683 Depends-On: https://review.opendev.org/c/starlingx/config/+/770135 Signed-off-by: Litao Gao Change-Id: Ib8a8a1844d5a151c15454f446129642ed4f4dfe8 commit 69664edc58bbf062d0628b0d283e66104f8d0a0e Author: Litao Gao Date: Mon Jan 11 08:51:24 2021 -0500 VF rate limiting support This feature is part of 'Single NIC support'. This commit implements the rate limit on VFs in terms of max_tx_rate leveraging hardware NIC driver capability. And adjust sriov-device-plugin config.json to make it easy to allocate rate limited VFs to the target Pods. Story: 2008470 Task: 41508 Depends-On: https://review.opendev.org/c/starlingx/config/+/770132 Signed-off-by: Litao Gao Change-Id: I3b609296b6b5b872f0bb0bd9733740b01e9d421c commit 52677deb2c56038d108edbd5ae7dd788b9501f77 Author: David Sullivan Date: Wed Jan 20 11:31:22 2021 -0600 Remove upgrade flags during abort If the upgrade flags are present during controller-1 downgrade, the controller will fail to unlock. Previously these were removed as part of the host-downgrade command, but this requires root access. These flags will now be removed by the conductor as part of the upgrade-abort command. Story: 2008055 Task: 41633 Signed-off-by: David Sullivan Change-Id: I7bb60e42d956140e06e44145ab8b168d1f3b72be commit a875f32e57b05420d3bc018bc14718a95591ba53 Author: David Sullivan Date: Sat Jan 23 13:16:22 2021 -0600 Remove host manifest during downgrade During the host-downgrade action the host manifest needs to be removed. This is similar to the process taken during host-reinstall. The manifest needs to be removed to prevent the host from running kubeadm prematurely. Story: 2008055 Task: 41674 Signed-off-by: David Sullivan Change-Id: I2242e8ab95228549ea3c1b3f6924025ae66ecb0f commit ac284046e449b3998db861b16db59dabbe8eb7bc Author: albailey Date: Mon Jan 25 14:27:52 2021 -0600 Fix tox failure when certain tmp folder un-writable When /tmp/device_images folder is not writeable, tox fails for three device_image API unit tests. This change mocks open and fdopen when calling that POST API endpoint, to allow the unit tests to pass. Closes-Bug: 1911997 Signed-off-by: albailey Change-Id: Ib6132a9b95b214e60ab34faa8a9615e1a2381fef commit c991e1e122db066f386c8ba4295f206fbae818e5 Author: Bin Qian Date: Wed Jan 20 14:23:47 2021 -0500 cert-mon secret data migration for upgrade to stx5 Add data migration code to populate secret data for cert-mon service. The secret data is stored in static secret data file. The data in static secret is only configured in initial bootstrap. Closes-bug: 1913173 Change-Id: I9ddb1aca9b2ba136facf1b3c294a273010e2a26b Signed-off-by: Bin Qian commit c0fadef2c1aa8a0335b69a27d6c41db757a78131 Author: Litao Gao Date: Thu Jan 7 01:43:37 2021 -0500 Single NIC support This feature is to allow user to configure CaaS and Data networks to single VF capable NIC during ‘Host interface configuration’ before unlock operation. This commit introduce a new subinterface type which is used to support for ethernet subinterface creation over pci-sriov class ethernet interface. Story: 2008470 Task: 41505 Signed-off-by: Litao Gao Change-Id: I2c148d24c73854e59ebb2be122e97017c6235c89 commit 6d4d5e384727548a68d1509d94fe8afc22b16e53 Author: David Sullivan Date: Wed Jan 20 10:25:33 2021 -0600 Migrate etcd after both controllers are upgraded The etcd database can go out of sync if we swact back and forth between controllers running different versions. This can happen if we need to abort the upgrade after swacting to controller-1. Since we are not upversioning etcd in this release we will address this by waiting to migrate the etcd data when both controllers are running the new release. The migration will now take place during the swact to controller-0 before upgrade-activate. This solution will present some problems when we do upgrade etcd, so further development will be required at that time. Story: 2008055 Task: 41630 Signed-off-by: David Sullivan Change-Id: I02b82bfe1a4b4b69aaa85d5f0d20246b9cda5629 commit 81df87b5b675c34d3216b5057153f510fcfa9baf Author: Sabeel Ansari Date: Fri Jan 22 17:48:28 2021 -0500 Fix urlparse call statement During previous commit of cert-mon, import urlparse statement was changed, which introduced a bug. This fixes the bug. Closes-Bug: 1911057 Change-Id: Ib90c89c3681ea97d77b5cdad4298cd7a08b42345 Signed-off-by: Sabeel Ansari commit 44d4774372d422d95ae198e778c5ddee32992f5a Author: Cole Walker Date: Fri Jan 22 16:55:10 2021 -0500 Add required vars for ptp-notification app Story: 2008529 Task: 41671 Signed-off-by: Cole Walker Change-Id: I807b03203a7589e5b1a04885dab09476323c5293 commit 6a51c9420a01ad48c8c8eec4835d9ed01a83d36c Author: Yuxing Jiang Date: Thu Jan 14 09:37:11 2021 -0500 Upgrade: fix hieradata software version mismatch This commit disables updating puppet hiera data in runtime if the software load of a host is different from the active controller, to prevent the hiera data software version mismatch during an upgrade/rollback procedure. Tested with an upgrade and rollback an AIODX system. Closes-Bug: 1911457 Change-Id: I42a953c0a31b80f2536de9292ed8b17789f6328c Signed-off-by: Yuxing Jiang commit 3477d1e3ed3266c091756acc07128277c31ee6db Author: Jessica Castelino Date: Thu Nov 19 04:19:48 2020 -0500 Remove identity entry on upgrade Identity services are no longer a shared service in DC. This commit updates the i_system table on subclouds to remove identity from the shared service list. Change-Id: I9a93c1aa364413d77af60a67c70a16ecbd546356 Signed-off-by: Jessica Castelino Closes-Bug: 1904675 Depends-On: https://review.opendev.org/#/c/763127/ commit f2dbdddcd160b643ccdc89b3d7b8e98acb25f6db Author: Angie Wang Date: Wed Jan 20 15:42:06 2021 -0500 Add Armada pod ready check in sysinv application audit In rare cases, after controller-0 is up from the initial unlock, armada pod is not running and ready yet when uploading platform-integ-apps. This will cause application upload failed. For the sysinv managed applications, update to ensure Armada pod is running and ready before attempting to upload/apply applications. Change-Id: I176bd1bbdb2ecf6285bd680091812aac43ea0ae3 Closes-Bug: 1912520 Signed-off-by: Angie Wang commit 93a86019cf68de0069702e7cf0ab7776f10c015f Author: Dan Voiculeasa Date: Fri Nov 6 17:40:49 2020 +0200 Introduce lifecycle operator The goal of this commit is to: - move most of the app specific logic outside conductor - introduce an operator for this purpose - keep the same logic as before - keep compatibility: apps that don't implement the new operator behave in a default way (same way as before) Create a lifecycle operator. Move lifecycle hooks from manifest operator to lifecycle operator. Move rbd actions from manifest operator to lifecycle operator. Move app specific code outside perform_app_operation to be executed by the app through lifecycle hooks. Conductor: - add lifecycle_hook_info param to all rpc perform_app_operation - introduce lifecycle hooks in perform_app_operation - remove some (de)activate_plugins in perform_app_operation - spawn only conductor.manager.perform_app_operation: greenthread.spawn(self.perform_app_upload greenthread.spawn(self.perform_app_apply Rest controller: - use manual mode hook in all rpc perform_app_operation - add semantic check hook for rpc operations: apply, remove, delete, abort, update Tested: - throw failed to download images (stx-openstack, snmp) - auto-apply hook throws exception on ceph down (platform-integ-apps) - upload (platform-integ-apps, stx-openstack, nginx-ingress-controller, snmp) - apply (platform-integ-apps, stx-openstack, cert-manager, nginx-ingress-controller, snmp) - remove (platform-integ-apps, stx-openstack, cert-manager, nginx-ingress-controller, snmp) - delete (platform-integ-apps, stx-openstack, cert-manager, nginx-ingress-controller, oidc-auth-apps, snmp) - auto apply platform-integ-apps - auto upload cert-manager, platform-integ-apps, oidc-auth-apps - create and delete cinder-volume in stx-openstack - application update (snmp) Change-Id: Ic83fbd25d23ae34889cb288330ec448f920bda39 Depends-On: Ibe994411fee55c84fa86770fad5497040f13b78f Depends-On: I41858c831a4af564dbdf38934d51d34489bf8a9a Depends-On: I533f2bd41c5627f83a004ffbf4543dd3c26d06b7 Story: 2007960 Task: 40463 Task: 41291 Signed-off-by: Dan Voiculeasa commit ad1188203e29dfe7d106efe449b08f222d0ecf1a Author: Sabeel Ansari Date: Tue Dec 22 12:27:56 2020 -0500 Use deployment namespace for platform certificates Instead of using kube-system namespace, we want to use deployment namespace for platform certificates. This commit includes: - Creation of namespace during upgrades - Changing namespace that cert-mon monitors Depends-on: https://review.opendev.org/#/c/768241/ Change-Id: Ieac392ed6d560be77327dd9e713fae01b17fda04 Story: 2007361 Task: 41165 Signed-off-by: Sabeel Ansari commit 20a761a073f5034d2548ee5edb4b04de63edd462 Author: Angie Wang Date: Wed Jan 13 12:35:18 2021 -0500 Clean up helm and armada directories from old release Delete the /opt/platform/helm and /opt/platform/armada directories from the old release at the stage of completing upgrade. Tested: AIO-SX and AIO-DX upgrades Change-Id: Ie2a582877135ac387c20bae4df7a6a6f244a0c3e Closes-Bug: 1910801 Signed-off-by: Angie Wang commit 576bdca7406712d2465ff36e216970e3da87152a Author: Angie Wang Date: Tue Jan 12 13:23:18 2021 -0500 Ensure Armada lock is released If the apply/remove of an application is failed due to an exception of the Armada request or an abnormal exit of the Armada request, the Armada lock cannot be released by Armada which causes the subsequent re-apply of the application to fail as it cannot acquire the lock. Update to delete the lock if an exception occurs during Armada request or Armada request exits abnormally. Closes-Bug: 1911243 Change-Id: I6ab02474831152b5b2b7302f0799c48d05fef64d Signed-off-by: Angie Wang commit 2ab82b7262e4881be8ca6997dfcf99a8931911d6 Author: albailey Date: Wed Jan 6 19:18:09 2021 -0600 Support passing an ignore alarm list to kube upgrade start API Health utils support an ignore alarm list. The kube_upgrade API makes use of those commands. Story: 2008137 Task: 41559 Signed-off-by: albailey Change-Id: I19db852f2e87273551d8a30f4bab470afa420de2 commit 84936cf189215ca60b9210ce27783f6331a98679 Author: Zhipeng Liu Date: Mon Jan 11 22:53:55 2021 +0800 Fix etcd AIO-DX upgrade failure issue The data migration is failed after controller-1 is upgraded (system host-upgrade controller-1) which is due to the error “/opt/platform/puppet/20.12/hieradata/static.yaml doesn’t exist.” when migrating the hieradata. Closes-Bug: 1894870 Change-Id: I71a5ef6d43da13039d0607b2457bd5f561704dfe Signed-off-by: Zhipeng Liu commit efa5f521c30a0d9bce4a8b0381d633f6f32fb4a1 Author: Angie Wang Date: Mon Sep 28 11:25:50 2020 -0400 Configure SQL as helm storage backend Configmap is the default helmv2 storage backend to store release information but its 1MB resource limit prevents scaling up stx openstack worker nodes, so we want to use SQL as helm storage backend. To configure SQL backend, generate helm database hieradata that will be used in puppet to create helm database. The helm database password is stored in keyring which can be retrieved in ansible playbook to configure database connection address. System upgrade support: The helm DB is new in the release stx5.0, so a password is generated for helm user. Helm user and password are written into the hieradata of release stx5.0. For AIO-SX upgrade, helm DB is created when applying bootstrap puppet manifest during ansible upgrade playbook. For two controllers upgrade, helm DB is created when applying upgrade puppet manifest during controller-1 upgrade. A migration script is created to migrate helm releases from configmap to postgresql. Partial-Bug: 1887677 Depends-On: https://review.opendev.org/#/c/761642/ Change-Id: I2f4f414068af297b5f4a3792c061443b7d3bdb32 Signed-off-by: Angie Wang commit f9b6c1c0a648abcdd6cdbc5a040895b0a3d7d4c9 Author: Sabeel Ansari Date: Mon Jan 4 16:49:16 2021 -0500 Update certificate pem files for unit tests Tox/Zuul failed unit tests due to expiring certificate pem files. This commit updates the expired (&extends other certs) for a long duration (10yrs). Closes-Bug: 1909817 Change-Id: I85cfb1896ae62947fb868315ecf1ff6b53fd4e08 Signed-off-by: Sabeel Ansari commit b6d64bef3935a58ac7ad946822f29fe7ed0fc316 Author: Takamasa Takenaka Date: Thu Jan 7 11:05:23 2021 -0300 Support trap_server_port configurable Add parameter for trap_server_port to allow sysadmin to configure snmp trap server port number through user helm override. Story: 2008132 Task: 41547 Signed-off-by: Takamasa Takenaka Depends-On: https://review.opendev.org/769732 Change-Id: Iac33ae11412852b59f6375abccb81988e85040e1 commit 06ae123d180648eb7b35c12c591198160d646db9 Author: Takamasa Takenaka Date: Fri Dec 18 17:06:17 2020 -0300 Remove REST API for host-based snmp -Remove controller related to icommunity/itrapdest API -Remove object related to icommunity/itrapdest API -Remove endpoint related to icommunity/itrapdest API -Remove db accessor related to icommunity/itrapdest API -Remove tables icommunity/itrapdest from sysinv -Remove data related to icommunity/itrapdest from dcorch Story: 2008132 Task: 41380 Change-Id: Ib1811a1bc507ee88e9a551a055068a3cd9d73a61 Depends-On: https://review.opendev.org/765381 Signed-off-by: Takamasa Takenaka commit bdf1888386094ee4a61fb4c78d268e187b1c3677 Author: Zhipeng Liu Date: Sat Oct 31 00:54:25 2020 +0800 Enable etcd with security setting. After upgrade from StarlingX 4.0 to 5.0, etcd will still keep insecurity before sending "system upgrade-activate". During upgrade activate stage, it will create etcd server/client certs and distribute them to all controllers before restart etcd and kube-apiserver with security enabled. Upgrade test pass on both simplex and duplex. Closes-Bug: 1894870 Depends-on: https://review.opendev.org/#/c/760510/ Change-Id: I27733a881a267e61502b36627dcab4136de23e3f Signed-off-by: Zhipeng Liu commit f4dc4a0c54bfacc0f30cf7c5b52fcd1b4ee4e025 Author: Takamasa Takenaka Date: Wed Dec 2 14:51:43 2020 -0300 Remove puppet entry related to host-based snmp With the host-based SNMP removal: -Remove trap destination entry -Remove snmp config entry Story: 2008132 Task: 41349 Change-Id: Ibd0faafe2255f6d325b11ab3383dc5f08f181955 Depends-On: https://review.opendev.org/765381 Signed-off-by: Takamasa Takenaka commit 12a73ed1126b3577d6f5fe450b0cfb6c81d8aab4 Author: Andy Ning Date: Wed Dec 23 13:59:04 2020 -0500 Update local puppet cache during runtime manifests apply This change added support to update local puppet hieradata cache during runtime manifests apply. The local cached puppet hieradata is used by controller_config during DOR to configure controllers. Change-Id: Ia5748230854ba54a9d73c7d5b44fdcb87d7cfe95 Closes-Bug: 1904739 Signed-off-by: Andy Ning commit 97ad36382ae7abeba0cf1118f3e80683f2d6215a Author: albailey Date: Fri Jan 1 10:40:58 2021 -0600 Update expired unit test certificate The sysinv unit test certificate expired Dec 22, 2020 This extends the date to Dec 22 2021. The unit test header was updated to trigger zuul to verify the changes are valid. A Readme file has been provided which lists the expiry dates for all the pem files used for the unit tests. Information on how to update the pem files will be added to the readme when all the expiry dates are aligned. Change-Id: I63cdd865a1818756f9ec3f20ebd724d2ad2d5729 Partial-Bug: 1909817 Co-Authored-By: Sabeel Ansari Signed-off-by: albailey commit 760e103ee4bf1190d907aa3532797018f832f9d1 Author: XinxinShen Date: Tue Dec 8 08:51:14 2020 +0800 Update the PCI device section in the API document An error occurred while using the API interface (/v1/devices/{device_id}) in the API documentation. After analysis code, I found that the API interface should be /v1/pci_devices/{device_id}, and the URL of "Modifies a specific PCI device" also has this problem, so I also update this part of the API document. Closes-Bug:#1907074 Signed-off-by: XinxinShen Change-Id: I636b830f34f9a74e0173d4a53224ec3f0249bba2 commit 596742595fe9cef9359e08c1c668cc1629de0351 Author: YuehuiLei Date: Thu Dec 31 16:32:25 2020 +0800 API documentation error for partitions An error occurred when using the partitions part of the API documentation (POST /v1/ihosts/{host_id}/partitions). After analyzing the code, I found that the URLs set in the code are /v1/partitions. This is inconsistent with the API documentation, so update the partitions section of the API documentation. Closes-Bug: #1897231 Change-Id: I9fc61b86bdfc4cb12acfd4f4cd8ecf8853ab338e Signed-off-by: YuehuiLei commit 8f89ba2e0eca77ffd417e0e7f549546b073ffe3c Author: YuehuiLei Date: Wed Dec 30 11:21:03 2020 +0800 Document error for post /V1/ilvgs interface. When I was using the /v1/ilvgs POST API, I found that the request parameters are mandatory rather than optional. Closes-Bug: #1897866 Change-Id: Iffb01013b16e62e1f3d3deb08f2b75132d62fde7 Signed-off-by: YuehuiLei commit fd2f1b105d2e46a05a60ce4efe650e94a5d26ce7 Author: Pablo Bovina Date: Mon Dec 7 09:23:15 2020 -0500 Remove system CLI for Net-SNMP commands The following commands are removed from CLI shell. snmp-comm-add: Add a new SNMP community. snmp-comm-delete: Delete an SNMP community. snmp-comm-list: List community strings. snmp-comm-show: Show SNMP community attributes. snmp-trapdest-add: Create a new SNMP trap destination. snmp-trapdest-delete: Delete an SNMP trap destination. snmp-trapdest-list: List SNMP trap destinations. snmp-trapdest-show: Show a SNMP trap destination. All the Net-SNMP configuration is responsibility of starlingx/snmp-armada-app. Story: 2008132 Task: 41367 Change-Id: I649d91d7e43f8b471434be5b433b47e8189600ce Depends-On: https://review.opendev.org/766267 Signed-off-by: Pablo Bovina commit 41b5fa0b507d85e77912d790fabb854ab788461b Author: Andy Ning Date: Tue Dec 15 13:31:39 2020 -0500 keep and reuse ssl certificate Currently when https is disabled, the installed ssl certificate is removed from the system. The default self signed certificate is installed again once https is enabled. This change enhanced ssl certificate handling in that: - The very first time https is enabled, the default self signed certificate is installed not only in fs but also in sysinv. - When https is disabled, installed ssl/tpm certificate is no longer deleted. - When https is enabled, the existing ssl/tpm certificate will be used if there is one installed. Otherwise the default self signed certificate will be installed (this is the case that https is enabled for the very first time). Change-Id: Iaef7b4acc4badaab617c05dcbd6654ea3d1e126a Closes-Bug: 1908437 Signed-off-by: Andy Ning commit e7c99304ce5acfc97ac29fe4b12fd020aac88bb7 Author: Lu Yao Chen Date: Wed Dec 16 14:43:40 2020 -0500 Resolving unpacking-non-sequence error Pylint error dealt with by per line suppression, error caused no staticmethod that breaks. Story: 2007082 Task: 37996 Signed-off-by: Lu Yao Chen Change-Id: I8d56ba47ee58edb070e99c5240afac1fd7942d7a commit 7312b1bf1de72e47bdea9d628f070afa1f4df8a1 Author: albailey Date: Thu Dec 10 18:25:11 2020 -0600 Turn off the legacy pip resolver for sysinv The requirements that had conflicts have been updated. The local upper-constraints is no longer being used by tox since two constraints files with different values for the same requirements are not supported. Additional pylint error codes are being suppressed as pylint is now running in python3 and has some newer checks. Cleans up some linter warnings related to yaml. Change-Id: I9b5158d59b0791e49c3037a8e823b67fb30c8292 Related-Bug: #1907678 Signed-off-by: albailey commit b5999ed6d180aa499252e869c436e0f2abf8338b Author: Andy Ning Date: Tue Dec 15 10:40:51 2020 -0500 Enhance puppet hieradata copy during controller config This change enhanced the copying of puppet hieradata during controller config by using rsync to "sync" hieradata to a temp cache directory, then rename it to the final cache directory. This is more atomic and minimize the chance to have incomplete or corrupted hieradata. Change-Id: I062ea54507a377e73102f29f40babc3d4fc214a6 Closes-Bug: 1904739 Signed-off-by: Andy Ning commit 51f1c359ae36ae5bf0dae0b69ebb0ecdf1235a72 Author: Lu Yao Chen Date: Tue Dec 15 11:14:35 2020 -0500 Resolving no-value-for-parameters pylint error With sqlalchemy documentation the following: insert(), update() and delete() objects do not take in arguments despite pylint errors for E11020 when calling objects from the same sqlalchemy library other instances work without arguments. Using per line suppression for the instances that raise pylint errors Story: 2007082 Task: 37996 Signed-off-by: Lu Yao Chen Change-Id: I6639bbca3ef7f1a48c4e242523ee6e62e624cbce commit 83445495d95fba3fff16d8845e60a487a0ffc4af Author: Lu Yao Chen Date: Mon Dec 14 12:17:32 2020 -0500 Resolving bad-except error Pylint catches bad-except error, in api/controllers/v1/interface.py, this part of code was not reached, resolving by removing except statement. Story: 2007082 Task: 37996 Signed-off-by: Lu Yao Chen Change-Id: I60ae5b8622e96c43d8f28febb0405259aadad0c0 commit 202902f60bf3483ae898b6f97979565f0c418c07 Author: Tee Ngo Date: Wed Dec 9 16:34:11 2020 -0500 Consistent creation of registry secret In this commit, the default-registry-key is created from sysinv credentials during the apply of the very first platform app (nginx-ingress-controller) in kube-system namespace. This key can no longer be removed. It will be used to create registry secret for all helm-based applications in their namespaces upon applying the application. Tests in AIOSX: - Install the system. - Remove cert-manager app, prune containerd image cache to force repull of the images from the registry, reapply the app. - Remove platform-integ-apps, verify that the default-registry-key in kube-system ns is not removed, reapply platform-integ-apps - Apply oidc-auth-apps, remove the app, verify that the default-registry-key in kube-system ns is not removed. Closes-Bug: 1906337 Change-Id: I042b1bf81f44a8c5498661c0bfca219c7150c57a Signed-off-by: Tee Ngo commit dfe0de2a0b2aecf4e5ea5c5fbbf1a77db4996f8a Author: Jim Gauld Date: Tue Jul 14 14:06:47 2020 -0400 Support upgrade to Helm v3 with containerized armada This provides an upgrade script for containerized armada. This launches armada using Helm v3. This does cleanup of the previous tiller-deployment, serviceaccounts and clusterrolebindings. Story: 2007927 Task: 40357 Depends-on: https://review.opendev.org/#/c/741024/ Change-Id: I4b3ba29c5210bcad269ed8dd25e00acafa1c8bb4 Signed-off-by: Jim Gauld Signed-off-by: Angie Wang commit b339eb000517a2627658da170b8aec0f1c131d7f Author: Mihnea Saracin Date: Fri Dec 4 11:35:23 2020 +0200 Fix upgrades that use restore procedure The restore procedure was recently changed, and it has additional steps. Upgrades that are based on backup and restore procedure don't automate the `complete the restore` additional step. In this fix, we automate this step when we do a `system upgrade-complete` command. Closes-bug: 1906557 Change-Id: I93db335ff058f987ed5f10ecac1aa402fe82032f Signed-off-by: Mihnea Saracin commit 2b211cf44f8bccd091206103ab72680dfc2cfac2 Author: Takamasa Takenaka Date: Tue Nov 3 15:17:00 2020 -0300 Add variables for snmp in fm.conf Snmp trap client needs the following three variables to connect to snmp trap server. - trap_server_ip - trap_server_port - snmp_enabled This commit is to update sysinv puppet hieradata with the value of snmp_enabled based on whether the snmp app is applied. trap_server_ip and trap_server_port are fixed. snmp_enabled takes 1(True)/0(False) depends on snmp armada app is applied or not (1 when applied). Change-Id: I76f9ef3e9ca6a98c0cd37a72f7aaccd10353b26b Story: 2008132 Task: 41205 Signed-off-by: Takamasa Takenaka commit 045e6c69e822369266831850f50d80da9b86b22c Author: Martin, Chen Date: Fri Nov 6 15:50:00 2020 +0800 Remove rook-ceph app constants Change-Id: I72f2dc74d8891e04ff693bac455f7940ddae0fb1 Closes-Bug: 1896628 Signed-off-by: Martin, Chen commit 083a073332d1654ee65f9e608e2974cedee0902f Author: Robert Church Date: Mon Oct 19 00:41:40 2020 -0400 Add support for disk wiping without GPT formatting To support a Rook based Ceph cluster, provide an option when wiping disks to skip adding a partition table. This allows preparing disks on all hosts during provisioning from the active controller that will be formatted and run as OSDs when the Rook k8s application is applied. Change-Id: I93a9770b0d78ddddc01fe7956d7ad7058acc8e71 Story: 2005527 Task: 41127 Signed-off-by: Robert Church