HA does not work in case if cluster node is suffering from out of memory
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Won't Fix
|
High
|
Fuel Library (Deprecated) | ||
7.0.x |
Won't Fix
|
High
|
Fuel Library (Deprecated) |
Bug Description
[root@fuel ~]# fuel --fuel-version
api: '1.0'
astute_sha: f7cda2171b0b677
auth_required: true
build_id: 2015-02-07_20-50-01
build_number: '76'
feature_groups:
mirantis
fuellib_sha: 64f3ebe9fcbd18b
fuelmain_sha: c799e3a6d88289e
nailgun_sha: 2ef819732a3ee7a
ostf_sha: 3b57985d4d21555
production: docker
release: 6.0.1
release_versions:
2014.2-6.0.1:
VERSION:
api: '1.0'
astute_sha: f7cda2171b0b677
build_id: 2015-02-07_20-50-01
build_number: '76'
feature_groups:
mirantis
fuellib_sha: 64f3ebe9fcbd18b
fuelmain_sha: c799e3a6d88289e
nailgun_sha: 2ef819732a3ee7a
ostf_sha: 3b57985d4d21555
production: docker
release: 6.0.1
Baremetal,Ubuntu, HA, Neutron-
Controllers:3 Computes:97
deployment was successful
rally light was succssfulhttp:
But after rally ull test I have found the following bihaviour on the primary controller:
root@node-32:~# ls -la
-bash: fork: Cannot allocate memory
root@node-32:~# top
-bash: fork: Cannot allocate memory
root@node-32:~# less /var/log/messages
-bash: fork: Cannot allocate memory
root@node-32:~# dmesg
-bash: fork: Cannot allocate memory
I could execute the following command, but after that i lost ssh connection:
root@node-32:~# exec ls
ceph.bootstrap-
Connection to node-32 closed.
I couldn't open nw ssh connection. I couldn't reach bash of the node even ipmi. (screenshot is attached)
I've rebooted the node at Feb 15 17:39:33 UTC. from messages (I can see log files after rebooting):
<6>Feb 15 17:39:33 node-32 kernel: imklog 5.8.6, log source = /proc/kmsg started.
latest message before rebooting:
<134>Feb 14 13:55:54 node-32 haproxy[11311]: 192.168.0.34:52990 [14/Feb/
Now can reach ivia ssh and execute bash cmmands
from messages log:
Feb 15 18:39:49 node-32 kernel: [ 3650.131654] perf samples too long (2504 > 2500), lowering kernel.
The latest strings from keystone:
<15>Feb 14 13:55:54 node-32 keystone-all Auth token not in the request header. Will not build auth context.
<15>Feb 14 13:55:54 node-32 keystone-all arg_dict: {}
I have no idea how I can know why node was hanged
Diagnostic snapshot is here: https:/
Changed in mos: | |
importance: | Undecided → Critical |
Changed in mos: | |
status: | New → Confirmed |
Changed in fuel: | |
assignee: | nobody → Fuel Library Team (fuel-library) |
status: | New → Confirmed |
importance: | Undecided → High |
importance: | High → Medium |
Changed in fuel: | |
importance: | Wishlist → High |
tags: | removed: qa-agree-7.0 |
tags: | added: qa-agree-7.0 |
tags: | added: release-notes-done |
tags: | added: ha |
We will require an environment. Can you please collaborate on skype?