non-matching names for network_interface fails
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| kolla |
Confirmed
|
Critical
|
Unassigned | ||
Bug Description
In globals.conf
network_interface: "bond0"
neutron_
In inventory file, need to overide for nodes that only have one nic
[compute]
eselde10u01s01 network_
eselde02u02
eselde02u03
Deployment fails for node eselde10u01s01
cat /tmp/kolla.log | grep -e PLAY -e TASK -e ansible_bond0
....... log snipplet ...................
PLAY [ceph-mon,
TASK [setup] *******
TASK [common : include] *******
TASK [common : Ensuring config directories exist] *******
TASK [common : Copying over config.json files for services] *******
TASK [common : Copying over heka config files] *******
TASK [common : Registering common role has run] *******
TASK [common : Copying over heka custom config file] *******
TASK [common : Copying over cron logrotate config files] *******
TASK [common : Creating log volume] *******
TASK [common : Starting heka bootstrap container] *******
TASK [common : Starting heka container] *******
TASK [common : Starting kolla-toolbox container] *******
TASK [common : Initializing toolbox container using normal user] ***************
TASK [common : Starting cron container] *******
TASK [common : Registering common role has run] *******
TASK [nova : include] *******
TASK [nova : Ensuring config directory exists] *******
TASK [nova : Copying over config(s)] *******
TASK [nova : Creating ceph erasure profile] *******
TASK [nova : Creating ceph ruleset] *******
TASK [nova : Creating ceph ruleset] *******
TASK [nova : Creating ceph pool] *******
TASK [nova : Creating ceph ruleset for cache] *******
TASK [nova : Creating ceph pool for cache] *******
TASK [nova : Adding cache to pool] *******
TASK [nova : Setting cache-mode] *******
TASK [nova : Setting cache overlay for pool] *******
TASK [nova : Setting cache hit_set_type] *******
TASK [nova : Pulling cephx keyring for nova] *******
TASK [nova : Pulling cephx keyring for libvirt] *******
TASK [nova : Pushing cephx keyring for nova] *******
TASK [nova : Pushing secrets xml for libvirt] *******
TASK [nova : Pushing secrets key for libvirt] *******
TASK [nova : Ensuring config directory exists] *******
TASK [nova : Find keyring files] *******
TASK [nova : Copy over ceph keyring file] *******
TASK [nova : Copy over ceph.conf] *******
TASK [nova : Pushing secrets xml for libvirt] *******
TASK [nova : Extract key from file] *******
TASK [nova : Pushing secrets key for libvirt] *******
TASK [nova : Creating the Nova service and endpoint] *******
TASK [nova : Creating the Nova project, user, and role] *******
TASK [nova : Setting sysctl values] *******
TASK [nova : Ensuring config directories exist] *******
TASK [nova : Copying over config.json files for services] *******
TASK [nova : Copying over nova.conf] *******
fatal: [eselde10u01s01]: FAILED! => {"failed": true, "msg": "'dict object' has no attribute u'ansible_bond0'"}
TASK [nova : Copying over libvirt configuration] *******
| Changed in kolla: | |
| milestone: | none → newton-rc2 |
| status: | New → Triaged |
| Steven Dake (sdake) wrote | #1 |
| Changed in kolla: | |
| status: | Triaged → Confirmed |
| importance: | Undecided → Critical |
| Jeffrey Zhang (jeffrey4l) wrote | #2 |
This is a bug in ansible fact caching. Fact caching records state of the system and keeps it around. Ansible is supposed to keep it around for 5 minutes. It seems like ansible is keeping is around forever.
Delete your ansible fact cache.
We need a documentation change to explain this problem - ~10 people in irc have reported it.