Yeah.  There would have been a lot of connection issues because I was doing a lot of testing and reconfiguring.    The only part that's really applicable for this issue is the period you mentioned from 11:54 to 12:01.  I did use the service vdsm status command after the host came back up and the service was not running.  I start the service manually and it comes up without error and the node comes back online.  Is it normal operation for the host to automatically recover if it can, including starting vdsmd?  One of my colleagues thinks that, perhaps that we are experiencing normal operation.  But I can't imagine that the host wouldn't come back completely if it's able.

On Tue, Jan 27, 2015 at 3:05 PM, Piotr Kliczewski <piotr.kliczewski@gmail.com> wrote:
Looking at the logs I can see that connection was lost at 2015-01-26
09:24:43,213
and I can see good number of reconnection attempts which end up with
timeout or 'no route to host'.
The connection was recovered at 2015-01-26 09:28:56,292.

Vdsm.log do not contain above connection loss (it starts at 2015-01-26
10:01:02,208).

It was lost again at 2015-01-26 11:54:58,741 and it was recovered at
2015-01-26 12:01:47,752.

I checked vdsm logs and I can see really weird lack of logs:

JsonRpc (StompReactor)::DEBUG::2015-01-26
11:52:35,893::stompReactor::98::Broker.StompAdapter::(handle_frame)
Handling message <StompFMainThread::INFO::2015-01-26
12:01:45,183::vdsm::131::vds::(run) (PID: 7021) I am the actual vdsm
4.16.10-8.gitc937927.el6 love005.ovt.visionamics.com
(2.6.32-504.3.3.el6.x86_64)
MainThread::DEBUG::2015-01-26
12:01:45,184::resourceManager::421::Storage.ResourceManager::(registerNamespace)
Registering namespace 'Storage'

which covers having no connection from the engine perspective.

Usually when there are connectivity issues we see timeouts in the logs
but here there are 'no route to host' as well
which suggest networking issues.

@Dan - Do you know what caused lack of logs in vdsm?
@ILanit - What vdsm version do you use?

On Tue, Jan 27, 2015 at 4:57 PM, Piotr Kliczewski <pkliczew@redhat.com> wrote:
>
>
>
>
> ----- Original Message -----
>> From: "Eli Mesika" <emesika@redhat.com>
>> To: "Piotr Kliczewski" <pkliczew@redhat.com>
>> Cc: "Artyom Lukianov" <alukiano@redhat.com>, users@ovirt.org, rabshear@citytwist.net, "ILanit Stein"
>> <istein@redhat.com>
>> Sent: Tuesday, January 27, 2015 4:39:26 PM
>> Subject: Re: [ovirt-users] Host remains Non-Responsive after reboot
>>
>>
>>
>> ----- Original Message -----
>> > From: "ILanit Stein" <istein@redhat.com>
>> > To: "Artyom Lukianov" <alukiano@redhat.com>, "Eli Mesika"
>> > <emesika@redhat.com>
>> > Cc: users@ovirt.org, rabshear@citytwist.net
>> > Sent: Tuesday, January 27, 2015 5:19:12 PM
>> > Subject: Fwd: [ovirt-users] Host remains Non-Responsive after reboot
>> >
>> >
>> > Hi Guys,
>> >
>> > Can you please look into this please?
>>
>> Hi
>> From the logs I can see clearly that host is turned on in 2015-01-26
>> 11:56:51,191
>> However, there is a stomp exception in 2015-01-26 11:56:53,544 and a
>> connection timeout in 2015-01-26 11:56:53,553 that might be related
>>
>> Piotr, can you please have a look ?
>>
>
> Sure. Can you please send me the logs?
>
>>
>> >
>> > Thanks,
>> > Ilanit.
>> > ----- Forwarded Message -----
>> > From: "Rob Abshear" <rabshear@citytwist.net>
>> > To: "ILanit Stein" <istein@redhat.com>
>> > Sent: Tuesday, January 27, 2015 3:05:56 PM
>> > Subject: Re: [ovirt-users] Host remains Non-Responsive after reboot
>> >
>> > Here are the logs. you requested.  The shutdown of the node was at 11:53
>> > and vdsmd was manually restarted at 12:01 to get the node back online.
>> >
>> > On Tue, Jan 27, 2015 at 2:05 AM, ILanit Stein <istein@redhat.com> wrote:
>> >
>> > > It might be a bug,
>> > > Would you please attach the logs, I mentioned bellow,
>> > > that can bring more details on the failure?
>> > > Adding Eli, that may want to give some input on this issue.
>> > >
>> > > Thanks,
>> > > Ilanit.
>> > >
>> > > ----- Original Message -----
>> > > From: "Rob Abshear" <rabshear@citytwist.net>
>> > > To: "ILanit Stein" <istein@redhat.com>
>> > > Cc: users@ovirt.org
>> > > Sent: Monday, January 26, 2015 9:43:14 PM
>> > > Subject: Re: [ovirt-users] Host remains Non-Responsive after reboot
>> > >
>> > > I have done a bit more investigating on this matter.  If I restart the
>> > > node
>> > > from within oVirt using the power management option "restart", then the
>> > > node restarts and vdsmd DOES NOT start.  If I go into the DRAC and issue
>> > > the command to power cycle the machine, then the machine restarts and
>> > > vdsmd
>> > > DOES start.  I can run the following command from another node in the
>> > > cluster:
>> > > fence_drac5 -a 192.168.200.105 -l root -p <password> -x -o reboot
>> > > and the node restarts and vdsmd DOES start.
>> > >
>> > > On Sun, Jan 25, 2015 at 1:56 AM, ILanit Stein <istein@redhat.com> wrote:
>> > >
>> > > > Hi Rob,
>> > > >
>> > > > Thanks for this report.
>> > > >
>> > > > Would you please provide these logs, at the time frame, the host
>> > > > failure
>> > > > occur:
>> > > > 1. oVirt Engine: /var/log/ovirt-engine/engine.log
>> > > > 2. host: /var/log/vdsm/vdsm.log
>> > > >
>> > > > If it is reproducible, please add this info as well.
>> > > >
>> > > > You can also check vdsm service status, on host, while host reported as
>> > > > Non responsive,
>> > > > by running on host 'service vdsmd status'
>> > > > There might some problem, that might have prevented from vdsm service
>> > > > to
>> > > > come up, on host.
>> > > >
>> > > > Ilanit.
>> > > >
>> > > > ----- Original Message -----
>> > > > From: "Rob Abshear" <rabshear@citytwist.net>
>> > > > To: users@ovirt.org
>> > > > Sent: Friday, January 23, 2015 9:22:42 PM
>> > > > Subject: [ovirt-users] Host remains Non-Responsive after reboot
>> > > >
>> > > >
>> > > > I am running oVirt Engine Version 3.5.0.1-1.el6. I have 4 hosts in the
>> > > > cluster. Each host has a drac5 and it is configured and working. I am
>> > > > trying to simulate a node failure. I am running one HA VM on one of the
>> > > > hosts for testing. I simulate the failure by powering off the host with
>> > > the
>> > > > VM running.
>> > > >
>> > > > Here is what is happening.
>> > > >
>> > > >
>> > > >     * Host is powered off
>> > > >     * ~4 minutes pass and the host is recognized as not responding
>> > > >     * Automatic fence runs and the VM migrates. Another host in the
>> > > >     node
>> > > > is chosen as a proxy to execute Status command on the host.
>> > > >     * Same host is chosen as proxy to execute Start command on the
>> > > >     host.
>> > > >     * Same host is chosen as proxy to execute Status command on the
>> > > >     host.
>> > > >     * The host DOES physically start.
>> > > >     * The host never shows status of UP.
>> > > >     * I select “confirm host has been rebooted” and I see a manual
>> > > >     fence
>> > > > start.
>> > > >     * Host stays non-responsive.
>> > > >     * I put the host in maintenance and then activate it.
>> > > >     * Host still non-responsive
>> > > >     * I put the host in maintenance and do a reinstall
>> > > >     * Reinstall finishes and host becomes UP
>> > > >
>> > > > So, everything seems to go fine with the HA functionality, but the host
>> > > > never recovers without being reinstalled. Please let me know which logs
>> > > you
>> > > > need to look at to help me out with this.
>> > > >
>> > > > Thanks
>> > > >
>> > > >
>> > > >         Sent with Mixmax
>> > > >
>> > > > _______________________________________________
>> > > > Users mailing list
>> > > > Users@ovirt.org
>> > > > http://lists.ovirt.org/mailman/listinfo/users
>> > > >
>> > >
>> >
>>
> _______________________________________________
> Users mailing list
> Users@ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users