Error while creating slaves #43
Comments
|
That exception is thrown by jenkins master when a jenkins slave is removed by the plugin when its idle. The unknown taskid is interesting. More logs around the task would help diagnose this. |
|
Indeed, I think the task ID is the actual problem. Here is a log with: grep -B 3 -A 3 mesos-jenkins-640f87cc-3092-4ccc-a117-0113df8f2139 /var/log/jenkins/jenkins.log > jenkins.log The log around the actual error looks like: |
|
Hi. I confirm the issue seems to be bound to the "Unknown taskId" message. It happened again and again I had the same message. |
|
Anything interesting in this thread? I'm having the same issue |
|
Looks like the scheduler is not running. Do you know why? Is it registered with the master? |
|
Well, Mesos thinks that it is running, but for whatever reason mesos-plugin thinks that it's not? |
|
Hello It seems that Mesos framework try to create the slave twice, resulting in a crash of the Mesos driver ... I have set in bold relevant information. Oct 29, 2014 10:45:06 AM org.jenkinsci.plugins.mesos.MesosComputerLauncher launch Oct 29, 2014 10:45:06 AM org.jenkinsci.plugins.mesos.JenkinsScheduler terminateJenkinsSlave Oct 29, 2014 10:45:06 AM org.jenkinsci.plugins.mesos.MesosComputerLauncher launch |
|
I also faced this issue and that is because of a race condition on scheduler start-up when there are multiple builds in queue. This pull request should address this issue: #70 |
|
Thanks @maselvaraj for the PR. I'm not exactly clear what the race is here. Can you add more explanation here or in your PR? Also, I realized that throwing an IllegalStateException on receiving status update from unknown task id is a bug. I will fix that shortly. |
|
Sure Vinod. I have added more details on the PR. Please let me know if you have any questions |
|
I am seeing the same symptoms as described in this old bug report with the latest version of the jenkins-mesos plugin (0.13.1) with mesos v1.0.1 and jenkins v2.7.4. I don't get the same stack traces but the unknown task id log bits along with ultimately the jenkins framework deregistering from mesos and being unable to create any new slaves for builds. An interesting addition to the description for me is that if we go into the Jenkins UI and save the global config settings, jenkins will then re-register with mesos and begin doing work again |
|
I'm having exactly the same issues as @sedninja: Jenkins 1.651.3 Mesos 1.1.0 and 0.13.1 (and also a self-built 0.13.1 with the mesos library dependency increased to 1.1.0). |
|
bump. Seeing this with version 0.14 of the plugin, mesos 1.0.1 - that's the version of my running cluster and the installed native lib. Specifically: |
…43) Summary: * Use only MesosPodRecordRepository, thus drop InMemoryPodRepository * Incorporate imperative USI interface
Hi,
I've a working mesos + jenkins setup. I can execute mesos jobs just fine most of the time.
However every now and then it looks like something happens in the jenkins plugin and it cannot create new slaves on demand anymore. Looking at the logs I see a few:
and at some point I get:
and the only solution is to restart the jenkins instance. Notice that running
mesos-executeby hand still works. Ideas? I'm using 0.18.2 on RHEL6.The text was updated successfully, but these errors were encountered: