+1 to improving the CI. I'm unable to help with either of your questions, though.
Thanks for pushing this forward. Michael Marshall On Thu, Mar 25, 2021 at 7:14 AM Lari Hotari <lari.hot...@sagire.fi> wrote: > Hi all, > > Pulsar CI build queues are long and there is a very small number of GitHub > Runner VMs processing build jobs. > > I don't have extensive statistics, but whenever I have gone to the "in > progress" listing of the build queue in the last recent days, it has been > never higher than 25. > For example, now there's only 3 jobs in progress shown at > https://github.com/apache/pulsar/actions?query=is%3Ain_progress . > However, the total length of the build queue was 83 at the same time, shown > at > https://github.com/apache/pulsar/actions?query=is%3Aqueued > > When you setup a "personal Pulsar CI" for your own forked repository, you > get 20 GitHub Runner VMs, all dedicated to your own builds. (see mailing > list thread "Setting up personal Pulsar CI for reducing load on > apache/pulsar CI") > A "personal Pulsar CI" is currently much more powerful than the > apache/pulsar CI. > > Why is there such a low amount of resources for apache/pulsar builds? The > issue https://github.com/apache/pulsar/issues/9154 mentions: > "Note: all Apache projects share the same limit of shared GitHub Actions > runners, and based on the chart below Pulsar is consuming 30+ runners while > the total limit (for all ASF projects) is 180." > > 1) Is some other ASF project consuming the shared quota or has there been a > change in the resource allocation for apache/pulsar? > 2) Where can we get the statistics for ASF and apache/pulsar GitHub Actions > resource consumption? > > I hope we get Pulsar CI in better shape soon. It seems that everyone else > is satisfied? :) > > BR, Lari >
