[Midonet] Some resources went into error state after upload_core_repos task was executed

[Oleksandr Balenko]

Steps to reproduce:
1. Install midonet plugin 3.0-3.0.1-1 on fuel master node v. 7.0 and deploy cluster with it.
2. Execute next commands on fuel master with deployed env:
2.1 fuel-createmirror -M
2.2 fuel --env <ENV_ID> node --node-id <NODE_ID1> <NODE_ID2> <NODE_ID_N> --tasks upload_core_repos
3. Wait few minutes.
4. Execute "crm status" on controller node

Actual Result:
Failed actions:
    p_mysql_monitor_60000 on node-18.test.domain.local 'unknown error' (1): call=152, status=complete, last-rc-change='Wed Apr 13 13:38:48 2016', queued=0ms, exec=0ms
    p_neutron-dhcp-agent_monitor_20000 on node-19.test.domain.local 'not running' (7): call=306, status=complete, last-rc-change='Thu Apr 14 08:34:42 2016', queued=0ms, exec=0ms
    p_mysql_start_0 on node-20.test.domain.local 'unknown error' (1): call=335, status=Timed Out, last-rc-change='Thu Apr 14 09:25:52 2016', queued=0ms, exec=300008ms
    p_neutron-dhcp-agent_monitor_20000 on node-20.test.domain.local 'not running' (7): call=302, status=complete, last-rc-change='Thu Apr 14 08:34:58 2016', queued=0ms, exec=0ms

Expected Result:
No failed actions should be displayed.

[Samir Ibradžić (sibradzic)]

@ Oleksandr,

I still haven't been able to reproduce this. Tried in very simple environment, 1xNSDB, 1xController & 1xCompute. What kind of node setup were you using?

[Samir Ibradžić (sibradzic)]

I tried to reproduce it again, this time with Controllers in HA setup and multiple Computes, to no avail. Please supply the log files of a failure on failed environment, as well as the versions package being upgraded (for example, output of *apt-cache policy neutron-server* and *dpkg-query -l neutron-server* before and after the upload_core_repos attempt)

[Oleksandr Balenko (obalenko)]

@Samir,

I suppose that I used HA setup while observed this issue.
But anyway, let's consider current issue as "incomplete", build new rpm with fix for "uninstall" issue, and I'll test both of them.
Because as for now it's really time consuming to resetup cluster one more time for providing logs, version of packages etc...
In case of repeated issue on the new rpm we can easily setup debug session. ok?

[Oleksandr Balenko (obalenko)]

Could reproduce this issue on rpm
830e944ee87b5d328f95363e3b1d40a5 midonet-fuel-plugin-3.0-3.0.1-1.noarch.rpm

[Oleksandr Balenko (obalenko)]

^^ update to previous comment:
couldn't reproduce

[Irina Povolotskaya (ipovolotskaya)]

Should we mark the bug as 'fix committed' then?

[Oleksandr Balenko (obalenko)]

Marked as "invalid".