Periodic Job Stalling

Hey all - great product with CircleCI and all.

Lately I’ve been running into issues where, on a Linux machine executor, the job stops reporting information back to the circle UI. I’ve seen evidence that, sometimes, this happens and the current job-task actually does finish!.. but not always.

I suspect my builds put an unexpected amount of memory pressure on the system, and this could be a result of that (I originally tried to keep these builds in the normal docker setup but it keeps hitting memory limits and killing processes in ways hard to compensate for), but I figured I should report something in case its more interesting.

The pipeline is github/robertfmurdock/Coupling


I have noticed the same issue with jobs not reporting information back to the web UI as well on Docker Linux Small and Windows Medium VM instances. Sometimes the UI receives no updates for 30+ seconds. The issue is quite intermittent, and I suspect it may be related to load from resource sharing, so possibly memory or other pressure but more likely from other customers than our own jobs. 100% of our jobs finish.