I suggest to halt work on production dc until we move at least a few hypervisors to use the vdsm scratch pad hook for local disk and migrate thier vms to use it, so we'll see a significant improvement in storage performance before moving on with production dc.
[ https://ovirt-jira.atlassian.net/browse/OVIRT-609?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17600#comment-17600 ]
Evgheni Dereveanchin commented on OVIRT-609:
--------------------------------------------
Here are some relevant messages from engine.log:
{quote}
grep 1394b752 /var/log/ovirt-engine/engine.log
2016-06-23 09:06:34,099 INFO [org.ovirt.engine.core.bll.CreateAllSnapshotsFromVmCommand] (ajp--127.0.0.1-8702-1) [1394b752] Lock Acquired to object EngineLock [exclusiveLocks= key: e7a7b735-0310-4f88-9ed9-4fed85835a01 value: VM
2016-06-23 09:06:35,708 INFO [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (org.ovirt.thread.pool-8-thread-15) Correlation ID: 1394b752, Job ID: a8fab0bf-d45e-46eb-8314-e22db8e6a3f4, Call Stack: null, Custom Event ID: -1, Message: Snapshot 'ngoldin_before_cluster_move' creation for VM 'jenkins-phx-ovirt-org' was initiated by admin.
2016-06-23 09:09:46,038 WARN [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (org.ovirt.thread.pool-8-thread-14) Correlation ID: 1394b752, Job ID: a8fab0bf-d45e-46eb-8314-e22db8e6a3f4, Call Stack: org.ovirt.engine.core.common.errors.VdcBLLException: VdcBLLException: org.ovirt.engine.core.vdsbroker.vdsbroker.VDSNetworkException: java.util.concurrent.TimeoutException (Failed with error VDS_NETWORK_ERROR and code 5022)
2016-06-23 09:09:47,859 ERROR [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (org.ovirt.thread.pool-8-thread-14) Correlation ID: 1394b752, Job ID: a8fab0bf-d45e-46eb-8314-e22db8e6a3f4, Call Stack: org.ovirt.engine.core.common.errors.VdcBLLException: VdcBLLException: org.ovirt.engine.core.vdsbroker.vdsbroker.VDSNetworkException: java.util.concurrent.TimeoutException (Failed with error VDS_NETWORK_ERROR and code 5022){quote}
Looks like VDSM was slow to respond (probably due to storage slowness) while the snapshot is likely to have completed fine. I'll review host logs and share my findings.
> Jenkins snapshot creation failed
> --------------------------------
>
> Key: OVIRT-609
> URL: https://ovirt-jira.atlassian.net/browse/OVIRT-609
> Project: oVirt - virtualization made easy
> Issue Type: Bug
> Reporter: Evgheni Dereveanchin
> Assignee: infra
>
> [~ngoldin@redhat.com] issued a live snapshot creation on the Jenkins VM to prepare it for cluster move. This failed and it's not really clear why. Relevant event logs below, suggesting that the hypervisor started dumping VM memory to the snapshot which caused a storage slowdown.
> {quote}2016-Jun-23, 18:06 Snapshot 'ngoldin_before_cluster_move' creation for VM 'jenkins-phx-ovirt-org' was initiated by admin.
> 2016-Jun-23, 18:09 Failed to create live snapshot 'ngoldin_before_cluster_move' for VM 'jenkins-phx-ovirt-org'. VM restart is recommended. Note that using the created snapshot might cause data inconsistency.
> 2016-Jun-23, 18:13 Host ovirt-srv02 has network interface which exceeded the defined threshold [95%] (em1: transmit rate[100%], receive rate [0%])
> 2016-Jun-23, 18:13 Storage domain Production experienced a high latency of 18.7802 seconds from host ovirt-srv11. This may cause performance and functional issues. Please consult your Storage Administrator.{quote}
--
This message was sent by Atlassian JIRA
(v1000.98.4#100004)
_______________________________________________
Infra mailing list
Infra@ovirt.org
http://lists.ovirt.org/mailman/listinfo/infra