pgj commented on PR #4958: URL: https://github.com/apache/couchdb/pull/4958#issuecomment-1914662730
> I was mainly suggesting a way to have a bound on the uncertainty in the stats, if it the increase in messages is a problem. @nickva I am afraid we cannot allow any uncertainty in the statistics if downstream consumers wanted to build uses cases on top of this mechanism that involves billing. > I don't know if it's a performance issue, I think you and @chewbranca did some measurements? @chewbranca might have some performance results on his side, but I have not collected any myself. Probably I should do some just for curiosity and for including here. > From what I remember in mango the issue was happening when we have a `limit` specified? Yes, but if I recall correctly there is always a limit assigned for the `_find` requests. I believe the emphasis is on the relative distance of the limit and the number of results returned. It is technically a result of a race condition: if the `stop` signal arrives at the wrong time, either the `completed` message is not emitted or it is not handled in the time. > Ideally we would wait for all workers to return their final message (perhaps as their exit reason as already noted), but without forcing the user to wait for more responses than are needed to form a complete response (i.e, the first response from each shard range). This can be done easily: replace the `stop` for an `ok` in `mango_cursor_view:handle_doc/2`. Actually, this is also case that could be tested for performance. Maybe it is not that expensive and sending the stats for every row. > Put simply, it was not anticipated that we would care about the other workers when this was built, the code fires off the jobs and wants to return as soon as it can, killing the remaining processes. A good fix here should tackle that directly. @rnewson do you have any pointers or ideas where to start with this? In the previous discussions, you had [some suggestions](https://github.com/apache/couchdb/pull/4735#issuecomment-1715743087), but I could not make them work. -- 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: notifications-unsubscr...@couchdb.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org