Samual stated: Ø the master does not bring an additional node online for a newly triggered job if there is already a node online which matches that label
I have the same general setup using VMware slaves that boot as required. I’m on Jenkins 1.656 and about a dozen slaves, with the git plugin and 2 monitor plugins at back revs due to some bugs. I use a generic label on each job that matches the label on multiple matching slaves. Except for four bugs, this works as expected and fires up another slave when the slave that it previously ran on is busy. E.g: if I want to build in a Docker image, I run it on any one of 4 slaves that have the “Dockerhost” label. There are 4 annoying bugs that are not fixed at this point: 1. After a network outage of a master node reboot, the VMware slaves do not fire up automatically. I have to manually bring them online once before things work as expected. 2. It can take up to about 5 minutes for Jenkins to notice that a slave is bootable, and actually do so. 3. The slave node monitor plugins (all of them that I’ve tried) have no concept of how to handle a node that is down because it is idle. The all mark this as a critical issue when it is actually expected behaviour. 4. If you take a slave manually offline to debug something, after a few minutes it will mark it as unreachable instead of intentionally offlined. This causes some unexpected monitoring issues as a consequence. The slave status checking is obviously in the wrong place in the scheduling logic. From: jenkinsci-users@googlegroups.com [mailto:jenkinsci-users@googlegroups.com] On Behalf Of samuel.e.bro...@gmail.com Sent: July-19-16 16:30 To: Jenkins Users Subject: Nodes Not Coming Back Online The system with which I am working uses nodes that are brought online as needed, and taken offline if they are not immediately used again (within two minutes) when a job is complete. Using the “restrict where this job can be run” or the Parameterized Build (using a label), the master does not bring an additional node online for a newly triggered job if there is already a node online which matches that label. This is not the desired behavior. Right now, this blocks the build queue by making the new job wait until the job already running on the online node is finished (multiple hours in some cases). I need the master to bring a node (with matching label) online for a new job unless there is already a FREE node online matching that label. Jenkins knows these jobs are waiting too, because the logs show "Launching machine XXXXX because it has been in demand for 1 hr 28 min". -- 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<mailto:jenkinsci-users+unsubscr...@googlegroups.com>. To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-users/89cf70ae-f350-479e-b2f3-dafb20322061%40googlegroups.com<https://groups.google.com/d/msgid/jenkinsci-users/89cf70ae-f350-479e-b2f3-dafb20322061%40googlegroups.com?utm_medium=email&utm_source=footer>. For more options, visit https://groups.google.com/d/optout. -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-users/ec0dcc67e393490cb3659d95c1c0b052%40mbx02cmb01p.esentire.local. For more options, visit https://groups.google.com/d/optout.