rnewson commented on PR #4729: URL: https://github.com/apache/couchdb/pull/4729#issuecomment-1686649525
I'm not entirely giving up on tcp keepalive (it ought to kill the process/port for the socket, which I think will kill the mochiweb process, but subject to testing) However, I'm beginning to think what we need is application level. There's clearly a modest, finite number for how long we're prepared to wait for a response to start, which Cloudant, for example, already expresses in haproxy `timeout server`. So I think couchdb needs a similar setting, that if a response hasn't begun in that timeframe, we exit (and fabric_streams cleanup then triggers). The only endpoint that delays the _start_ of the response by a potentially very long time is `_find`, because it can discard work. A filtered changes response has at least started the response even if nothing passes the filter. -- 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]
