You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Version: What redis-py and what redis version is the issue happening on?
redis-py 4.4.2 / redis 6.0.5
Platform: What platform / version? (For example Python 3.5.1 on Windows 7 / Ubuntu 15.10 / Azure)
Python 3.9 on MacOS Ventura 13.4.1
Description: Description of your issue, stack traces from errors and code that reproduces the issue
Recently, I've faced a situation where connections never get returned to a connection pool, when there is a surge of Redis commands. while investigating the incident, i've found out that the below scenario can cause a connection leak
ClusterPipeline is stacked with commands that needs to be directed to at least two nodes
let's say node 1 and node 2
ClusterPipeline obtains a connection to node 1 successfully
ClusterPipeline fails to obtain a connection to node 2 due to ConnectionError
ClusterPipeline retries the entire process again, but doesn't release an acquired connection to node 1 before doing that
below code includes a test case to cause a connection leak and a possible solution to the problem
Hi @zakaf! First off, thank you for the report, and a potential bug fix.
@dvora-h mind checking if this makes sense in the 4.6, and 5.0 contexts since that's getting much closer to release? Then we can figure out what we may want to do here.
Uh oh!
There was an error while loading. Please reload this page.
Version: What redis-py and what redis version is the issue happening on?
redis-py 4.4.2 / redis 6.0.5
Platform: What platform / version? (For example Python 3.5.1 on Windows 7 / Ubuntu 15.10 / Azure)
Python 3.9 on MacOS Ventura 13.4.1
Description: Description of your issue, stack traces from errors and code that reproduces the issue
Recently, I've faced a situation where connections never get returned to a connection pool, when there is a surge of Redis commands. while investigating the incident, i've found out that the below scenario can cause a connection leak
below code includes a test case to cause a connection leak and a possible solution to the problem
Let me know, if you guys think this is an actual issue :)
The text was updated successfully, but these errors were encountered: