[Users] [vdsm] Remove and Add host does not work

[Douglas Landgraf]

On 02/28/2012 08:49 AM, Deepak C Shetty wrote:
> On 02/28/2012 07:11 PM, Itamar Heim wrote:
>> On 02/28/2012 03:31 PM, Deepak C Shetty wrote:
>>> On 02/28/2012 06:57 PM, Itamar Heim wrote:
>>>> On 02/28/2012 03:22 PM, Deepak C Shetty wrote:
>>>>> On 02/28/2012 06:46 PM, Itamar Heim wrote:
>>>>>> On 02/28/2012 02:48 PM, Deepak C Shetty wrote:
>>>>>>> Hi,
>>>>>>> I had a host managed via OE, completely working fine. Was even 
>>>>>>> able to
>>>>>>> create and run VMs off it.
>>>>>>> I tried removing the host from the OE (put host into maint. mode 
>>>>>>> and
>>>>>>> then remove) and when i re-discover
>>>>>>> the same host, OE just keeps seeing it as Non-responsive.
>>>>>>
>>>>>> what do you mean by "re-discover"?
>>>>>>
>>>>>
>>>>> I just meant that i removed the host and re-added it by selecting 
>>>>> "New"
>>>>> on the Hosts tab
>>>>> and putting the IP and hostname.
>>>>>
>>>>>>>
>>>>>>> I am able to ssh into the host and see that none of the vdsm 
>>>>>>> processes
>>>>>>> have been started.
>>>>>>> I tried doing "Confirm host has been rebooted" on the OE, did not
>>>>>>> help.
>>>>>>> I tried putting
>>>>>>> host into maint. mode and re-activating the host, just doesn't help
>>>>>>
>>>>>> did you do any change to the host?
>>>>>> just removing it from engine shouldn't cause vdsm to know/care and
>>>>>> should work just like before.
>>>>>>
>>>>>
>>>>> Nothing changed on the host. In fact when i removed and added the
>>>>> host back
>>>>> it says everything is installed so does nothing but reboots the host,
>>>>> post reboot
>>>>> vdsm does not start automatically.... and host status is 
>>>>> non-responsive
>>>>> on OE
>>>>
>>>> if vdsm does not start, OE is correct...
>>>> does vdsm try to start and fails (and if so, log excerpt?)
>>>
>>> Right, but I dont see any traces of vdsm trying to start. Nothing in 
>>> the
>>> vdsm.log
>>> thats relevant to this. I just did chkconfig --list and don't see vdsmd
>>> in that, could
>>> that be the reason.
>>>
>>> On the host I manually did `service vdsmd restart` and then everythign
>>> works
>>> fine.
>>>>
>>>>
>>>
>>
>> you are basically saying "re-installing a host causes vdsm to not 
>> start by default".
>> reproducing this again to make sure and opening a bug seems the right 
>> course.
>> trying to trace the install flow to provide root cause or even a 
>> patch would help more to fix this
>>
> Are you saying that vdsmd service should start by default post reboot 
> and its entry
> should be listed as part of chkconfig --list ?
>
vdsm uses systemd.
http://fedoraproject.org/wiki/SysVinit_to_Systemd_Cheatsheet

Try to use chkconfig vdsm on, you will see a wrapper to systemd command.

> Found this examining the vds bootstrap complete py log
>
> 2012-02-28 23:45:32,022 DEBUG    deployUtil 707 _updateFileLine: 
> return: True
> 2012-02-28 23:45:32,022 DEBUG    deployUtil 228 setVdsConf: ended.
> 2012-02-28 23:45:32,022 DEBUG    deployUtil 103 ['/bin/systemctl', 
> 'reconfigure', 'vdsmd.service']
> 2012-02-28 23:45:32,026 DEBUG    deployUtil 107
> 2012-02-28 23:45:32,026 DEBUG    deployUtil 108 Unknown operation 
> reconfigure
>
> 2012-02-28 23:45:32,026 DEBUG    deployUtil 103 ['/sbin/reboot']
> 2012-02-28 23:45:32,325 DEBUG    deployUtil 107
>
>
Which vdsm version are you using? If I am not wrong, I remember to see a 
patch for this report.. /me going to check..

Thanks!

-- 
Cheers
Douglas