You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On taking a jvm heap dump at multiple times during the day for a high qps finagle server, we see that the connections set/map is not flushing out some of the older connections that just stay and keep taking up large amounts of memory
Steps to reproduce the behavior
A high qps server with some huge responses(think 50mb)
The text was updated successfully, but these errors were encountered:
Hi @vikramkadi, have you tried reproducing this issue with a more recent version of Finagle? The 6.39 version appears to be from 2016 and the code has changed significantly. I would recommend trying an upgrade and seeing if the issue persists. If it is still there on the latest release, we will need to gather some more context to assist. Thank you.
Finagle http server(version: 6.39.0) seems to not release older connections from the concurrenthashmap/set increasing old gen space overtime.
Expected behavior
The connections bookkeeping set/map should clean up older entries(https://github.com/twitter/finagle/blob/finagle-6.39.0/finagle-core/src/main/scala/com/twitter/finagle/server/StackServer.scala#L230).
Actual behavior
On taking a jvm heap dump at multiple times during the day for a high qps finagle server, we see that the connections set/map is not flushing out some of the older connections that just stay and keep taking up large amounts of memory
Steps to reproduce the behavior
The text was updated successfully, but these errors were encountered: