jonasplaum opened a new issue, #5470:
URL: https://github.com/apache/couchdb/issues/5470
## Description
Every two or three days a single request to Nouveau fails with
`500 Internal Server Error
{"error":"unknown_error","reason":"connection_closed"}`
We checked with TCP Dumps between CouchDB and Nouveau.
It looks like one of the 16 Request (we have 16 shards) gets no http
response from nouveau. The last request over this TCP connection was almost
exactly 30 seconds ago. This would cause the connection close from dropwizard.
Maybe its a race condition, and couchdb sends a request while nouveau closes
the connection. This would explain, why we only get this once in a while and
only for one request.
CouchDB could send keep alive packets, or ibrowse needs to close unused
connections in under 30 seconds (or maybe increase the dropwizard timeout).
Then there is no risk for using connections that are already closed, or get
closed while couchdb sends a request.
For now we will increase the idleTimeout in nouveau, till a fix is available.
## Expected Behaviour
Nouveau requests should not fail randomly.
## Your Environment
CouchDB 3.4.2 on Debian 12 using the provided packages for Debain.
--
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]
