Replies: 6 comments 3 replies
-
You could check your log contents and the size of the logs (maybe also log-rotate if used). |
Beta Was this translation helpful? Give feedback.
-
@skeyby I went through the same issue a while back. It looks like couchdb is trying to keep the indexes 'warm'. So it, for whatever reasons, keeps indexing non-stop. Check here. You could try to reduce the batch_channels, but I have personally set all the _design/docs - to autoupdate: false. If there is any better fix to that I'm looking forward to it. Until then I will hold any other production update to 3.* |
Beta Was this translation helpful? Give feedback.
-
@skeyby do you have any .compact files around that were created with 2.1? If yes, this is it: #3001 You can safely delete the .compact files (don’t miss the ones for view indexes), and things should compact normally and then things should slow down. |
Beta Was this translation helpful? Give feedback.
-
This still is a big issue. Deleting .compact files - does not seem to fix it. Any other suggestion? |
Beta Was this translation helpful? Give feedback.
-
Mb something from https://docs.couchdb.org/en/stable/maintenance/performance.html could help? |
Beta Was this translation helpful? Give feedback.
-
In our case, it was bad coding practice. There was a while loop in our code was constantly running while True:
changes = db.changes(since=since, include_docs=True, heartbeat=60000)
# do stuff with changes (if any)... We ended up adding sleep in the while for a quick patch, I don't know if there is a better way to fix it. while True:
time.sleep(sleep_time) # <- added
changes = db.changes(since=since, include_docs=True, heartbeat=60000)
# do stuff with changes (if any)... |
Beta Was this translation helpful? Give feedback.
-
Hello,
I have a machine that keeps constantly at 100%. It used to do it with CouchDB 2.1 and now, after an upgrade to CouchDB 3.1, it keeps doing it.
The Tasklist show practically nothing running and even placing the log in debug mode reports nothing interesting.
The machine is mostly idle and just have a local replication job running (from a local db to another local one).
How can I understand what's going on?
I'm attaching the output for "top" from the machine, with thread details open... any suggestion's welcome.
Beta Was this translation helpful? Give feedback.
All reactions