[ovirt-users] Slow booting host - restart loop

[Bernardo Juanicó]

Hi Eli,

I could not access the psql prompt, i tried with the credentials and values
on /etc/ovirt-engine/engine.conf.d/10-setup-database.conf.

I tried interactively and also with a .pgpass file containing:
localhost:5432:engine:engine:PASSWORD

And i get the following error:

psql: FATAL:  Peer authentication failed for user "engine"

Thanks!

Bernardo


PGP Key <http://pgp.mit.edu/pks/lookup?op=get&search=0x695E5BCE34263F5B>
Skype: mattraken

2017-09-05 12:14 GMT-03:00 Eli Mesika <emesika at redhat.com>:

> Hi Bernardo
>
> I would like to suggest a workaround to this problem , can you please
> check that :
>
> We have a configuration value named FenceQuietTimeBetweenOperationsInSec.
> It controls the minimal timeout to wait between fence operation (stop,
> start),
> currently, it is defaulted to 180 sec , The key is not exposed to
> engine-config, so, I would suggest to
>
> 1) Change this key value to 900 by running the following from psql prompt :
>
> update vdc_options set option_value = '900' where option_name = '
> FenceQuietTimeBetweenOperationsInSec';
>
> 2) Restart the engine
>
> 3) Repeat the scenario
>
> Now, the engine will require 15 min between fencing operations and your
> host can be up again without being fenced again.
>
> Please let me know if this workaround is working for you
>
> Thanks
>
> Eli
>
> On Tue, Sep 5, 2017 at 4:20 PM, Bernardo Juanicó <bjuanico at gmail.com>
> wrote:
>
>> Martin, thanks for your reply, i was aware of the [1] BUG and the
>> implemented solution, changing ServerRebootTimeout to 1200 didnt change a
>> thing...
>> Now i know about [2] and ill test the fix once it gets released.
>>
>> Regards,
>>
>> Bernardo
>>
>> PGP Key <http://pgp.mit.edu/pks/lookup?op=get&search=0x695E5BCE34263F5B>
>> Skype: mattraken
>>
>> 2017-09-05 8:23 GMT-03:00 Martin Perina <mperina at redhat.com>:
>>
>>> Hi Bernardo,
>>>
>>> we have added timeout to wait until host is booted [1] in oVirt 4.1.2.
>>> This timeout is by default 5 minutes, but it can be extended using
>>> following command:
>>>
>>>    engine-config -s ServerRebootTimeout=NNN
>>>
>>> where NNN is number of seconds you want to wait until host is booted up.
>>>
>>> But be aware that you may be affected by [2], which we are currently
>>> trying to fix.
>>>
>>> Regards
>>>
>>> Martin Perina
>>>
>>>
>>> [1] https://bugzilla.redhat.com/show_bug.cgi?id=1423657
>>> [2] https://bugzilla.redhat.com/show_bug.cgi?id=1477700
>>>
>>>
>>> On Fri, Sep 1, 2017 at 7:54 PM, Bernardo Juanicó <bjuanico at gmail.com>
>>> wrote:
>>>
>>>> Hi everyone,
>>>>
>>>> I installed 2 hosts on a new cluster and the servers take a really long
>>>> to boot up (about 8 minutes).
>>>>
>>>> When a host crashes or is powered off the ovirt-manager starts it via
>>>> power management, since the servers takes all that time to boot up the
>>>> ovirt-manager thinks it failed to start and proceeds to reboot it, several
>>>> times before giving up, when the server is finally started (about 20
>>>> minutes after the failure)
>>>>
>>>> I changed some engine variables with engine-config trying to set a
>>>> higher timeout, but the problem persists.
>>>>
>>>> Any ideas??
>>>>
>>>>
>>>> Regards,
>>>> Bernardo
>>>>
>>>>
>>>> PGP Key
>>>> <http://pgp.mit.edu/pks/lookup?op=get&search=0x695E5BCE34263F5B>
>>>> Skype: mattraken
>>>>
>>>> _______________________________________________
>>>> Users mailing list
>>>> Users at ovirt.org
>>>> http://lists.ovirt.org/mailman/listinfo/users
>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20170905/2fa14a4f/attachment.html>