jonasplaum commented on issue #5887: URL: https://github.com/apache/couchdb/issues/5887#issuecomment-3909021927
Thanks for the Analysis here and in Slack. We will Upgrade to 3.5.1 and check if its happening again. It may take some time, because we have a bunch of CouchDB and nouveau servers running for some time and it was the first time this happend. Just an info (maybe its useful, otherwise feel free to ignore): I think the amount of running search queries may be a symptom. We can see that we got around 400 running queries in the time frame where nouveau doesn't work. In normal times on this System there are between 2 and 5 running at a time. <img width="1181" height="490" alt="Image" src="https://github.com/user-attachments/assets/9333af68-3cee-4b22-b4bc-b6ac9a4f5e72"; /> I think this was a buildup, till we hit an other timeout of 60 Minutes. And it looks like its climbing till 400 and stays there, maybe there is a limit reached. <img width="1464" height="569" alt="Image" src="https://github.com/user-attachments/assets/c2aa8da6-7eea-4fe8-a64c-0058e0c5cce4"; /> This area with around 60 Minutes per search request ends exactly after the restart of the erlang side of nouveau. The amount of requests does not change. We can also se, that the JVM restart is not visible here. I would expect that if we reached some kind of limit, that a restart would close the connections and free limited resources. We also checked our application logs and didn't find more requests than usual going to CouchDB / Nouveau. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
