I have occasionally a python worker using clickhouse-driver client that is stuck forever.
I have yet to really pinpoint exactly where but I suspect an issue with this client because there is nothing else that could hang forever other than a call made with the client (I have logs before / after).
I cannot provide how to reproduce the problem (I wish I could) as it occurs sometimes once every several thousands of queries but digging into the client looking for blocking calls I noticed this method:
Should not be the culprit since I've yet to see a PartiallyConsumedQueryError (and we try to acquire with blocking=False) but shouldn't the lock be released in case of PartiallyConsumedQueryError?
Not exactly sure what is the point of that lock anyway since we do not even look if we did acquire the lock or not.
I'm not seeing any other usage for that log than this method.
Otherwise if someone has any idea where there could be blocking calls without the client I'd be interested.
Describe the bug
Hi ๐๐ป
I have occasionally a python worker using clickhouse-driver client that is stuck forever.
I have yet to really pinpoint exactly where but I suspect an issue with this client because there is nothing else that could hang forever other than a call made with the client (I have logs before / after).
I cannot provide how to reproduce the problem (I wish I could) as it occurs sometimes once every several thousands of queries but digging into the client looking for blocking calls I noticed this method:
Should not be the culprit since I've yet to see a
PartiallyConsumedQueryError
(and we try to acquire with blocking=False) but shouldn't the lock be released in case of PartiallyConsumedQueryError?Not exactly sure what is the point of that lock anyway since we do not even look if we did acquire the lock or not.
I'm not seeing any other usage for that log than this method.
Otherwise if someone has any idea where there could be blocking calls without the client I'd be interested.