rnewson commented on PR #4729:
URL: https://github.com/apache/couchdb/pull/4729#issuecomment-1685950963

   @nickva 😱 interesting!
   
   What we've observed (a huge rise in interactive ioq requests after a series 
of `_find` requests that haproxy classified as `sH--`, and the new 
`mango_stats` showing a mango process that exits some hours after those after 
having examined the same number of documents as the database has in total) can 
only currently be explained by these processes simply continuing despite the 
client having gone away (the client being haproxy here, it closes the 
connection after its own `timeout server` interval).
   
   I can confirm what you're seeing, and I have also done so for an unindexed 
`_find` request by adding timer:sleeps in various places of the remote side of 
fabric_view_all_docs.
   
   So I guess we're looking for a gap in the `fabric_streams` error handling 
rather than a complete absence of orphaned worker cleanup code.


-- 
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]

Reply via email to