This repository has been archived by the owner on Apr 26, 2024. It is now read-only.
-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
Federation readers regularly get stuck at 100% CPU #10698
Labels
X-Needs-Info
This issue is blocked awaiting information from the reporter
Comments
It seems completely wedged merging dicts. Python stack trace from GDB:
C stack trace:
|
Nothing was logged after it wedged, which suggests it was making zero forward progress. |
matrix.org is on commit fe3466a. Room was |
reivilibre
added
T-Defect
Bugs, crashes, hangs, security vulnerabilities, or other reported issues.
X-Needs-Info
This issue is blocked awaiting information from the reporter
and removed
T-Defect
Bugs, crashes, hangs, security vulnerabilities, or other reported issues.
labels
Aug 26, 2021
Federation readers 1 & 2 both experienced the same issue at roughly 10am UTC today. The latest logs from the processes: Federation reader 1
Federation reader 2
We restarted each which got them going again. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Description
Federation reader workers regularly get stuck at 100% CPU for not explicable reason (graphs don't reveal why it's spinning CPU). Investigation is needed into what it is doing, and why this is happening.
The impact of the readers being down generally doesn't affect the local server itself, but does affect every other server's ability to interact with that server, such as joining rooms, looking up profiles, inviting other members, etc. The failure ends up making the other servers look bad when the cause isn't their fault :(
Steps to reproduce
Unclear
Version information
The text was updated successfully, but these errors were encountered: