[#QB-2380] QuickBuild Node Issues

QuickBuild

QuickBuild Node Issues

Created: 20/Mar/15 08:10 PM Updated: 24/Mar/15 11:13 PM

Component/s:

None

Affects Version/s:

6.0.9

Fix Version/s:

None

Original Estimate:	Unknown	Remaining Estimate:	Unknown	Time Spent:	Unknown
Environment:	QuickBuild 6.0.9

Description

« Hide

Hi,
I am having two issue with the QuickBuild nodes as the follows:
1) I received many alerts with similar message, "Build Agent '<AgentName>' is removed from active list as server has not hear from it for 12 seconds" although the build machines were still up and running fine. Please advise me on this issue.
2) When a node is offline, all of the jobs which are assigned to that node are automatically cancelled by QuickBuild. I think the jobs are supposed to wait for the node until it is back online and then resume the operations. Please advise me on this issue too.

Thanks,
-Phong

All

Comments

Work Log

Change History

Sort Order:

[ Permlink | « Hide ]

Robin Shen [21/Mar/15 12:58 AM]

For issue1, the agent is actually removed but then it has been added back automatically when server hears from it again. You may edit system setting to set a large value for agent timeout so that server does not complain about this.
For issue2, QB has to terminate the build as long as it finds the agent offline as it does not know when the agent will come back. If it waits eternally, other jobs can be blocked.

[ Permlink | « Hide ]

Phong Trinh [21/Mar/15 04:28 AM]

For issue 1: Regarding editing system setting to set a large value for agent timeout, Is it the wrapper.ping.timeout=0 in the wrapper.conf on QuickBuild server?
For issue 2: I think when a node is offline, QB can cancel the build which is running on the node. But it should not cancel all of builds in queue which are waiting for the node.

[ Permlink | « Hide ]

Robin Shen [22/Mar/15 12:18 AM]

For issue 1: this setting can be found in "Administration / System Setting" via QB web UI.
For issue 2: to have queueing builds waiting instead of failed, please configure master step of them to use resource instead of matching a particular node, for instance, you may define a resource say "node1", which matches node1 via its resource setting, then configure the master step to use that resource.

[ Permlink | « Hide ]

Phong Trinh [24/Mar/15 06:50 PM]

Thank you very much for responding promptly.
The solution for issue 1 fixed my system. Thank you.
I have an agent which dedicates to one of our builds. The build run for hours, and I schedule it run twice day. Per your suggestion, I think I create a resource which has only one node and configure master step to use that resource. My follow-up question is that will the builds be queued or wait if the node is offline?

Thank you in advance,
ptrinh

[ Permlink | « Hide ]

Robin Shen [24/Mar/15 11:13 PM]

Yes it will. Please give it a try.