We have used a mongo v8.0.1 replicaSet (2 nodes + arbiter) on our production servers (Rocky 9.4 with 12 cores) since some weeks. I restartet the whole system of primary after RAM upgrade some days ago.
We have seen 100% cpu usage caused by mongo process on all cores since that.
We have tested folllowing steps:
mongo service restart
look for slow queries (We also tried to stop all our services calling mongo, but cpu usage won’t change.)
node reinitilized
switch primary to another node
disable transparent huge page on os level
According to mongo, the connection count of node slowly increases overtime. The result of
following statement:
db.currentOp({“secs_running”: {$gte: 3}})
contains a lot of hello commands.
Does anyone has an idea?
Any help would be greatly appreciated
Downgrading openssl to 3.0 rectifies the issue, per the latest comment it is addressed in an upcoming release:
About the fix - we’ve actually got it working in our master branch and have backported it to 8.0 (it’ll be in the upcoming 8.0.5 release). We’re still working on bringing it back to 7.0 - that should take a few more weeks.