[Upgrade] Add validation which checks if there is enough disk space for ugprade
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Released
|
Medium
|
Evgeniy L | ||
5.0.x |
Fix Released
|
Medium
|
Evgeniy L |
Bug Description
Steps to reproduce:
1. Deploy simple flat environmnet(Centos + nova Flat 1 controller + 1 compute)
2. Run OSTF to validate cluster
3. Download latest tarball (in my case 11)
4. unpack it to /home directory on master ndoe
5. Run upgrade script (with enabled docker and openstack options)
6. Upgrade fails on upgrade nginx container(by reason a small free space size)
7. Rallback starts (and it is really good)
Expected:
Rallback finishes with success, all new containers were removed, supervisor starts all old container, UI is available, cluster is operational
Actual Result:
Rallback hangs on step of stoping services using supervisor. (2014-06-23 16:03:39 INFO 17651 (supervisor_client) Stop all services
)
But actually services are stopped
[root@nailgun docker-logs]# supervisorctl status
dhcrelay_monitor BACKOFF Exited too quickly (process log may have details)
docker-astute STOPPED Jun 23 03:59 PM
docker-cobbler STOPPED Jun 23 03:59 PM
docker-mcollective STOPPED Jun 23 03:59 PM
docker-nailgun STOPPED Jun 23 03:59 PM
docker-nginx STOPPED Jun 23 04:00 PM
docker-ostf STOPPED Jun 23 03:59 PM
docker-postgres STOPPED Jun 23 03:59 PM
docker-rabbitmq STOPPED Jun 23 03:59 PM
docker-rsync STOPPED Jun 23 03:59 PM
docker-rsyslog STOPPED Jun 23 04:00 PM
Seems we should verify status of container and try to stop only those one that hase running state.
As result I waiting near one our, and my deployment is non functional anymore...
P.S. See logs for details :)
summary: |
- [Upgrade] Rallback fails if upgrades fails by reason to small free disk + [Upgrade] Rollback fails if upgrades fails by reason to small free disk space |
summary: |
- [Upgrade] Rollback fails if upgrades fails by reason to small free disk - space + [Upgrade] Add validation which checks if there is enough disk space for + ugprade |
Changed in fuel: | |
assignee: | Igor Kalnitsky (ikalnitsky) → Evgeniy L (rustyrobot) |
status: | New → Confirmed |
importance: | High → Medium |
no longer affects: | fuel/5.1.x |
Changed in fuel: | |
status: | Fix Committed → Fix Released |
May be we should check and|or notify user about available free space before start upgrades|rollback?