I noticed at point the iops limit was getting hit. Upgrading the storage specs could be a great idea too.
On Wed, Mar 18, 2020, 11:59 Patrick Mu <zm2...@columbia.edu> wrote: > Dear Community, > > Our developers have identified frequently occurrence of "Cannot contact > <slave-name>" issue > in our CI system. Sheng and Leonard have helped to investigate this and > have found the CI master's network bandwidth reaching limit is probably the > culprit of the issue. To remove the burden of repeated CI retriggering from > developers, we decided to take the following steps: > > 1) Stop the CI Jenkins master > 2) Resize the CI master instance to a larger instance for more network > bandwidth capacity > 3) Restart the master > > The workflow will take less than 1 hour to complete (ideally 5-10 mins). > > In the meanwhile, if you already have PRs currently running in the CI, > please resubmit your PRs to make sure they will run the pipeline after > restart. > > We are sorry for any inconvenience caused. > > Best Regards, > > Ziyi >
