We have a cleanup pipeline with a single stage with one job set to run on
all agents.
It's triggered by a Cron Timer Specification, so there is no automatic
pipeline scheduling.
With 11 agents, it will run 11 jobs, but it might run twice on some agents
and never
run on other agents! There is no obvious patterns in which agents are
repeated or
ignored. It varies between runs. Has anyone else come across this?
This seems to happen the whole time. It's as if it decides to run on 11
agents, and then
just gives the jobs to the first eleven agents who ask for work, regardless
of whether
they already ran this pipeline instance.
It can look like this. Note that jobs 1 and 2 runs on agent 004, jobs 3 and
9 on 007, and jobs 6 and 8 runs on 005.
defaultJob-runOnAll-1 Passed Completed build-go-agent004(x.x.x.70)
defaultJob-runOnAll-10 Passed Completed 5 minutes and 13 seconds
build-go-agent014(x.x.x.88)
defaultJob-runOnAll-11 Passed Completed 3 minutes and 20 seconds
build-go-agent013(x.x.x.86)
defaultJob-runOnAll-2 Passed Completed 11 seconds
build-go-agent004(x.x.x.70)
defaultJob-runOnAll-3 Passed Completed build-go-agent007(x.x.x.79)
defaultJob-runOnAll-4 Passed Completed build-go-agent009(x.x.x.82)
defaultJob-runOnAll-5 Passed Completed 1 minute and 23 seconds
build-go-agent001(x.x.x.76)
defaultJob-runOnAll-6 Passed Completed build-go-agent005(x.x.x.77)
defaultJob-runOnAll-7 Passed Completed 1 minute build-go-agent011(x.x.x.84)
defaultJob-runOnAll-8 Passed Completed 13 seconds
build-go-agent005(x.x.x.77)
defaultJob-runOnAll-9 Passed Completed 11 seconds
build-go-agent007(x.x.x.79)
I don't know why the column "duration" is sometimes empty, but that's
common also
for pipelines that seem to work.
We have other pipelines with run on all agents enabled, and I've never seen
this
problem on the others. (I didn't do an extensive search though.) One
pipeline even
has just the same config as I describe in the first paragraph. It just runs
an hour
earlier, and has a different <exec>, but it still seems to work.
The problematic pipeline's config looks like this:
<pipeline name="clean_old_dockers" isLocked="false">
<trackingtool ... />
<timer>0 0 22 ? * MON-FRI</timer>
<materials>
<pipeline ... />
</materials>
<stage name="clean">
<approval type="manual" />
<jobs>
<job name="defaultJob" runOnAllAgents="true">
<tasks>
<exec command="/home/go/.local/bin/clean_old_dockers">
<arg>-m</arg>
<runif status="passed" />
</exec>
</tasks>
</job>
</jobs>
</stage>
</pipeline>
--
You received this message because you are subscribed to the Google Groups
"go-cd" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
