1

I'm diagnosing why our Rails servers seem to leak memory.

In looking at New Relic today, I noticed that "Free heap slots" of our Ruby VMs seems to jump from almost nothing up to the baseline level of "Live objects" at seemingly random intervals.

My guess is that this has something to do with GC, but I'm not sure, and I'm not sure how to fix it.

The vertical lines are deployments that reap processes.

New Relic Graph

More detail, for reference (this is across 4 hosts each running 12 Unicorn worker processes):

enter image description here

enter image description here

Ross
  • 131
  • 5
  • Hey @Ross. Have you found an answer? – Sergey Alekseev Oct 16 '15 at 09:43
  • No, not really. We ended up using Unicorn Killer to keep the leaks from growing without bounds, but I never figured out what caused the leaks or this behavior in the first place. – Ross Oct 17 '15 at 02:09

0 Answers0