Force jenkins to refresh the scheduler label information
Hi, We use jenkins along with the libvirt plugin that automatically start up slaves when jobs containing matching labels are placed in the job queue. I assign labels dynamically to the slave images by means of a groovy script that monitors the build queue at specified intervals. This works but I have some issues that I do not think have anything to do with libvirt plugin but rather the way the jenkins scheduler works (I might be wrong). - When there are jobs in the queue with the label "pickMe" and I assign that label to the slave that I want libvirt to start, there is a delay of a couple of minutes before the slave is started. - If I manually start the slave just after the label is assigned, the job with label "pickMe" waits in the job queue for a couple of minutes (while there is an online slave with the label "pickMe"). - This gives me a feeling that the label data of the slaves are updated on a periodic interval. Is there any way I can manually force an update so that this delays will go away? Any ideas will be appreciated. Regards Hendrik -- 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.
Re: Concurrent builds on dumb slave configuration
The problem with this issue was that the number of executors of the master was limiting the execution on the slaves. Solution was to increase this number. On Monday, October 22, 2012 9:07:00 AM UTC+2, Hendrik Meiring wrote: > > Hi, > > We have been using Jenkins/hudson for a couple of years and I have now > converted my build project to make use of the concurrent launching of > projects. > In short we have a couple of projects that must be compiled on four > different os/architectures. > > *Node Setup* - unchanged from previous working setup. > Four dumb slaves running on four different VM's with the number of > executors set to 1. Their names are devel-win32, devel-win64, devel-lin32 > and devel-lin64. > > *Job setup* > The job is set up as a 'Build multi-configuration project' with the 'The > build is parameterized' option checked. The parameterized section contains > a Node parameter with my 4 nodes specified as possible nodes. > The 'Allow multi node selection for concurrent builds ' option is > selected and the 'Execute concurrent builds if necessary' is selected. > > I am aware of https://issues.jenkins-ci.org/browse/JENKINS-15370 that > make concurrent build triggering using multi-selection not an option for me > but now I am experiencing the following with the above setup: > > When I manually trigger the above job (single) selecting a different > platform/node each time the builds are correctly triggered and appears to > be running underneath the appropriate node on the left hand panel but after > more detailed investigating I found that one of the jobs are building and > the other jobs are blocked on the other machines until the job that is busy > are completed - thus sequential building.. grrr > > Am I missing something here? > > Thanks in advance > Hendrik >
Re: Concurrent builds on dumb slave configuration
Any input/advice on this issue? On Monday, October 22, 2012 9:07:00 AM UTC+2, Hendrik Meiring wrote: > > Hi, > > We have been using Jenkins/hudson for a couple of years and I have now > converted my build project to make use of the concurrent launching of > projects. > In short we have a couple of projects that must be compiled on four > different os/architectures. > > *Node Setup* - unchanged from previous working setup. > Four dumb slaves running on four different VM's with the number of > executors set to 1. Their names are devel-win32, devel-win64, devel-lin32 > and devel-lin64. > > *Job setup* > The job is set up as a 'Build multi-configuration project' with the 'The > build is parameterized' option checked. The parameterized section contains > a Node parameter with my 4 nodes specified as possible nodes. > The 'Allow multi node selection for concurrent builds ' option is > selected and the 'Execute concurrent builds if necessary' is selected. > > I am aware of https://issues.jenkins-ci.org/browse/JENKINS-15370 that > make concurrent build triggering using multi-selection not an option for me > but now I am experiencing the following with the above setup: > > When I manually trigger the above job (single) selecting a different > platform/node each time the builds are correctly triggered and appears to > be running underneath the appropriate node on the left hand panel but after > more detailed investigating I found that one of the jobs are building and > the other jobs are blocked on the other machines until the job that is busy > are completed - thus sequential building.. grrr > > Am I missing something here? > > Thanks in advance > Hendrik >
Re: Concurrent builds on dumb slave configuration
Any input on this? Am I misinterpreting how the system should work, am I configuring something wrong or is this a bug that must be logged?
Concurrent builds on dumb slave configuration
Hi, We have been using Jenkins/hudson for a couple of years and I have now converted my build project to make use of the concurrent launching of projects. In short we have a couple of projects that must be compiled on four different os/architectures. *Node Setup* - unchanged from previous working setup. Four dumb slaves running on four different VM's with the number of executors set to 1. Their names are devel-win32, devel-win64, devel-lin32 and devel-lin64. *Job setup* The job is set up as a 'Build multi-configuration project' with the 'The build is parameterized' option checked. The parameterized section contains a Node parameter with my 4 nodes specified as possible nodes. The 'Allow multi node selection for concurrent builds ' option is selected and the 'Execute concurrent builds if necessary' is selected. I am aware of https://issues.jenkins-ci.org/browse/JENKINS-15370 that make concurrent build triggering using multi-selection not an option for me but now I am experiencing the following with the above setup: When I manually trigger the above job (single) selecting a different platform/node each time the builds are correctly triggered and appears to be running underneath the appropriate node on the left hand panel but after more detailed investigating I found that one of the jobs are building and the other jobs are blocked on the other machines until the job that is busy are completed - thus sequential building.. grrr Am I missing something here? Thanks in advance Hendrik
