We have fallen behind a bit recently, but try to keep up with the LTS releases, just upgraded to 1.480.3. One weekend a month our IT department has a scheduled patch of the packages on the server OS and that is usually when we also arehitching a ride on that restart and do plugin upgrades at the same time, it is a bit risky to do those at the same time since if something goes wrong you have two places to look for, but since that big amount of jobs and build used to take over an hour for Jenkins to start up we felt the risk was worth it. After the 1.480.3 upgrade Jenkins restarts in only 20 minutes, so maybe we'll change that policy.
A few days before the scheduled restart we do our homework with some testing on our test servers and deeper analysis of some of the "high risk" plugins (reading changelogs etc.) to prepare for anything that might happen or if a plugin upgrade should be skipped until we can address any problems that occurred during testing. But trying to mimic our big production environment in the testing environment is close to impossible in terms of load etc. So that's why the deeper analysis is important so that we can have some kind of mitigation plan if any surprises comes up the days after the upgrade. Core upgrades are a bit more extensive in terms of testing and preparation, we try, when we have time, to help the community with the LTS testing and then we have a bunch of test cases targeted for our internal plugins and specific environment(s). The aim is to have upgraded to the new core a couple of weeks after LTS release, but that is not always the case, this time for example we upgraded to 1.480 about the same time as 1.509 was released :). That was mostly due to lack of time and some concerns around how much bigger load we would get on the cluster due to the parallel matrix builds feature. A wall of text, sorry for that, hope I didn't bore anyone :) I have submitted a talk proposal to JUC in October around this topic, let's hope it gets accepted and you'll get even more info then ;) Robert Sandell Software Tools Engineer - SW Environment and Product Configuration Sony Mobile Communications From: [email protected] [mailto:[email protected]] On Behalf Of martinda Sent: den 12 juni 2013 13:10 To: [email protected] Subject: Re: How many jobs can jenkins support? Robert, On Tuesday, June 11, 2013 10:50:44 AM UTC-4, Robert Sandell wrote: For one of our servers we have roughly 4K jobs and 250 slaves running about 5K builds per day, on average there is one executor per slave. What version of Jenkins are you using, and what is your upgrade policy (upgrade of plugins versions and upgrade of Jenkins versions)? Thanks, Martin -- You received this message because you are subscribed to the Google Groups "Jenkins Developers" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]<mailto:[email protected]>. For more options, visit https://groups.google.com/groups/opt_out. -- You received this message because you are subscribed to the Google Groups "Jenkins Developers" 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/groups/opt_out.
