Hi,

For our nightly testing, we revert slaves to a VMware snapshot before
starting jobs.

Most of the time this works, but sometimes the first job after the revert
job fails.

The weird part is that a downstream job that runs on the same slave
will often succeed less than two minutes after the upstream failure.

I'm guessing that Jenkins thinks the slave is still online and rejecting
the connection and/or not connecting.

My idea for a work-around is to add a "snooze" time after the revert.  This
way, Jenkins can figure out that the slave is offline.

So, what amount of time is necessary for Jenkins to realize that a slave is
offline?
5, 10, ?? minutes?

Note that for non-windows slaves, I'm not disconnecting the slave prior to
revert.  I had a "good reason" for doing this at the time, but can not
remember it.

Below are some snippets of console output showing my situation.

Running Jenkins ver. 1.509.1

Thanks,
Dale

------------------------------------
http://jenkins/job/Job-A/549/console
------------------------------------
21:45:38 NOT Disconnecting Hudson client before powering on machine:
AT03-cent5.  (non-Windows slave)
21:45:38 Revert vm 'AT03-cent5' to snapshot '2014-11-03'.
21:45:38 Power on vm AT03-cent5.
21:45:38 Spent 4.327958 seconds powering on.
21:45:38 Snooze an extra 180 seconds to allow boot completion...
<snip>
21:48:40 Finished: SUCCESS

-------------------------------------
http://jenkins/job/Job-B/1109/console
-------------------------------------
21:48:46 Started by upstream project "Job-A" build number 549
<snip>
21:48:49 Building remotely on AT03-cent5 in workspace
/opt/hudson/workspace/Job-A
21:51:28 FATAL: hudson.remoting.RequestAbortedException:
java.io.IOException: Unexpected termination of the channel
<snip>

----------------------------------------
http://jenkins/job/Job-C/740/consoleFull
----------------------------------------
21:52:51 Started by upstream project "Job-B" build number 1109
21:52:54 Building remotely on AT03-cent5 in workspace
/opt/hudson/workspace/Job-C
<snip>
22:35:13 Finished: SUCCESS


-- 
Dale Quigg
Revolution Analytics, Inc.

-- 
 

Revolution R Plus <http://revolutionanalytics.com/plus>

Subscribe to Technical Support & Indemnification for R

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to jenkinsci-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to