[ovirt-devel] suspend_resume_vm fail on master experimental

Francesco Romani fromani at redhat.com
Wed Jan 11 10:42:10 UTC 2017



On 01/11/2017 09:29 AM, Yaniv Kaul wrote:
>
>
> On Wed, Jan 11, 2017 at 10:26 AM, Francesco Romani <fromani at redhat.com
> <mailto:fromani at redhat.com>> wrote:
>
>     Hi all
>
>
>     On 01/11/2017 08:52 AM, Eyal Edri wrote:
>>     Adding Tomas from Virt.
>>
>>     On Tue, Jan 10, 2017 at 10:54 AM, Piotr Kliczewski
>>     <piotr.kliczewski at gmail.com <mailto:piotr.kliczewski at gmail.com>>
>>     wrote:
>>
>>         On Tue, Jan 10, 2017 at 9:29 AM, Daniel Belenky
>>         <dbelenky at redhat.com <mailto:dbelenky at redhat.com>> wrote:
>>         > Hi all,
>>         >
>>         > test-repo_ovirt_experimental_master (link to Jenkins) job
>>         failed on
>>         > basic_sanity scenario.
>>         > The job was triggered by
>>         https://gerrit.ovirt.org/#/c/69845/
>>         <https://gerrit.ovirt.org/#/c/69845/>
>>         >
>>         > From looking at the logs, it seems that the reason is VDSM.
>>         >
>>         > In the VDSM log, i see the following error:
>>         >
>>         > 2017-01-09 16:47:41,331 ERROR (JsonRpc (StompReactor))
>>         [vds.dispatcher] SSL
>>         > error receiving from <yajsonrpc.betterAsyncore.Dispatcher
>>         connected ('::1',
>>         > 34942, 0, 0) at 0x36b95f0>: unexpected eof (betterAsyncore:119)
>>
>
>     Daniel, could you please remind me the jenkins link? I see
>     something suspicious on the Vdsm log.
>
>
> Please use my live system:
> ssh mini at ykaul-mini.tlv.redhat.com
> <mailto:mini at ykaul-mini.tlv.redhat.com> (redacted)
> then run a console to the VM:
> lagocli --prefix-path /dev/shm/run/current shell engine 
>
> (or 'host0' for the host)
>
>     Most notably, Vdsm received SIGTERM. Is this expected and part of
>     the test?
>
>
> It's not.

I fooled myself. We have two hosts here. host0 was holding the VM up
until the suspend. Then Engine decided to resume the VM on the other
one, host1.
While the VM was resuming, host0 experienced network issues, which led
to soft-fencing. That explains the mess on host0, even though it is still
unclear why host0 had network issues and heartbeat exceeded in the first
place.

On host1 the waters are even darker. The VM was resumed ~02:36

2017-01-11 02:36:04,700-05 INFO 
[org.ovirt.engine.core.vdsbroker.CreateVmVDSCommand] (default task-17)
[72c41f12-649b-4833-8485-44e8d20d2b49] FINISH, CreateVmVDSCommand,
return: RestoringState, log id: 378da701
2017-01-11 02:36:04,700-05 INFO 
[org.ovirt.engine.core.bll.RunVmCommand] (default task-17)
[72c41f12-649b-4833-8485-44e8d20d2b49] Lock freed to object
'EngineLock:{exclusiveLocks='[42860011-acc3-44d6-9ddf-dea64
2caf083=<VM, ACTION_TYPE_FAILED_OBJECT_LOCKED>]', sharedLocks='null'}'
2017-01-11 02:36:04,704-05 INFO 
[org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector]
(default task-17) [72c41f12-649b-4833-8485-44e8d20d2b49] Correlation ID:
72c41f12-649b-4833-8485-44e8d20d2b
49, Job ID: a93b571e-aed1-40e7-8d71-831f646255fb, Call Stack: null,
Custom Event ID: -1, Message: VM vm0 was started by admin at internal-authz
(Host: lago-basic-suite-master-host1).

While well within the timeout limit (600s), the vm was still restoring
its state:

2017-01-11 02:37:31,059-05 DEBUG
[org.ovirt.engine.core.vdsbroker.vdsbroker.GetAllVmStatsVDSCommand]
(DefaultQuartzScheduler8) [5582058d] START,
GetAllVmStatsVDSCommand(HostName = lago-basic-suite-master-host1,
VdsIdVDSCommandParametersBase:{runAsync='true',
hostId='0336661b-721f-4c55-9327-8fd2fd3a0542'}), log id: 1803c51b
2017-01-11 02:37:31,059-05 DEBUG
[org.ovirt.vdsm.jsonrpc.client.reactors.stomp.impl.Message]
(DefaultQuartzScheduler8) [5582058d] SEND
destination:jms.topic.vdsm_requests
reply-to:jms.topic.vdsm_responses
content-length:103

{"jsonrpc":"2.0","method":"Host.getAllVmStats","params":{},"id":"f2997c1d-49cc-4d2e-937c-e4910fbb75df"}^@
2017-01-11 02:37:31,059-05 DEBUG
[org.ovirt.vdsm.jsonrpc.client.reactors.stomp.StompCommonClient]
(DefaultQuartzScheduler8) [5582058d] Message sent: SEND
destination:jms.topic.vdsm_requests
content-length:103
reply-to:jms.topic.vdsm_responses

<JsonRpcRequest id: "f2997c1d-49cc-4d2e-937c-e4910fbb75df", method:
Host.getAllVmStats, params: {}>
2017-01-11 02:37:31,062-05 DEBUG
[org.ovirt.vdsm.jsonrpc.client.reactors.stomp.impl.Message] (SSL Stomp
Reactor) [5e453618] MESSAGE
content-length:829
destination:jms.topic.vdsm_responses
content-type:application/json
subscription:fe930de2-aa67-4bc4-a34c-be22edd1623e

{"jsonrpc": "2.0", "id": "f2997c1d-49cc-4d2e-937c-e4910fbb75df",
"result": [{"username": "Unknown", "displayInfo": [{"tlsPort": "-1",
"ipAddress": "192.168.201.4", "type": "vnc", "port": "-1"}, {"tlsPort":
"-1", "ipAddress": "192.168.201.4", "type": "spice", "port": "-1"}],
"hash": "-4942054084956770103", "acpiEnable": "true", "displayIp":
"192.168.201.4", "guestFQDN": "", "vmId":
"42860011-acc3-44d6-9ddf-dea642caf083", "pid": "0", "timeOffset": "0",
"session": "Unknown", "displaySecurePort": "-1", "displayPort": "-1",
"memUsage": "0", "guestIPs": "", "kvmEnable": "true", "displayType":
"vnc", "cpuUser": "0.00", "monitorResponse": "0", "elapsedTime": "389",
"vmType": "kvm", "cpuSys": "0.00", "status": "Restoring state",
"guestCPUCount": -1, "appsList": [], "clientIp": "", "statusTime":
"4295562220", "vmName": "vm0"}]}^@
2017-01-11 02:37:31,062-05 DEBUG
[org.ovirt.vdsm.jsonrpc.client.internal.ResponseWorker] (ResponseWorker)
[] Message received: {"jsonrpc": "2.0", "id":
"f2997c1d-49cc-4d2e-937c-e4910fbb75df", "result": [{"username":
"Unknown", "displayInfo": [{"tlsPort": "-1", "ipAddress":
"192.168.201.4", "type": "vnc", "port": "-1"}, {"tlsPort": "-1",
"ipAddress": "192.168.201.4", "type": "spice", "port": "-1"}], "hash":
"-4942054084956770103", "acpiEnable": "true", "displayIp":
"192.168.201.4", "guestFQDN": "", "vmId":
"42860011-acc3-44d6-9ddf-dea642caf083", "pid": "0", "timeOffset": "0",
"session": "Unknown", "displaySecurePort": "-1", "displayPort": "-1",
"memUsage": "0", "guestIPs": "", "kvmEnable": "true", "displayType":
"vnc", "cpuUser": "0.00", "monitorResponse": "0", "elapsedTime": "389",
"vmType": "kvm", "cpuSys": "0.00", "status": "Restoring state",
"guestCPUCount": -1, "appsList": [], "clientIp": "", "statusTime":
"4295562220", "vmName": "vm0"}]}

Then a failed migration attempt, which fails

2017-01-11 02:46:07,826-05 INFO 
[org.ovirt.engine.core.bll.MigrateVmToServerCommand] (default task-4)
[27938262-f5fe-41ff-a662-5177a3a25ada] Lock Acquired to object
'EngineLock:{exclusiveLocks='[42860011-acc3-44d6-9ddf-dea642caf083=<VM,
ACTION_TYPE_FAILED_VM_IS_BEING_MIGRATED$VmName vm0>]', sharedLocks='null'}'
2017-01-11 02:46:07,827-05 WARN 
[org.ovirt.engine.core.bll.MigrateVmToServerCommand] (default task-4)
[27938262-f5fe-41ff-a662-5177a3a25ada] Validation of action
'MigrateVmToServer' failed for user admin at internal-authz. Reasons:
VAR__ACTION__MIGRATE,VAR__TYPE__VM,VAR__HOST_STATUS__UP,ACTION_TYPE_FAILED_VDS_STATUS_ILLEGAL
2017-01-11 02:46:07,827-05 INFO 
[org.ovirt.engine.core.bll.MigrateVmToServerCommand] (default task-4)
[27938262-f5fe-41ff-a662-5177a3a25ada] Lock freed to object
'EngineLock:{exclusiveLocks='[42860011-acc3-44d6-9ddf-dea642caf083=<VM,
ACTION_TYPE_FAILED_VM_IS_BEING_MIGRATED$VmName vm0>]', sharedLocks='null'}'

which makes sense, host0 was being reinstalled (!?) meantime.

After that I see no more evidence of 'vm0' in the logs. I couldn't yet
cross-correlate with the tests, perhaps the test failed meantime?

TL;DR: we need a reproducer to keep investigating this, so far I've seen
quite some suspicious activity but nothing really concrete.

PS: last few lago failures I've seen are been triggered by completely
unrelated patches, so far I thik they are sporadic failures


-- 
Francesco Romani
Red Hat Engineering Virtualization R & D
IRC: fromani

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/devel/attachments/20170111/6fe98119/attachment-0001.html>


More information about the Devel mailing list