Revisiting this bug while we're triaging stale bugs - we haven't recorded a recent CI run that reproduces this issue; however, i think going by our prior analysis the "share_type_destroy" method here could use a db retry on deadlock. If multiple quota operations are occurring at the same exact instant, with no retry capability, we're open to hitting a deadlock - we have retry decorators on many other "update"/"delete" operations in the database code. We should probably add one and see if this helps.
I'm going to set this to "low-hanging-fruit" to see if any new contributors can pick this up.
Revisiting this bug while we're triaging stale bugs - we haven't recorded a recent CI run that reproduces this issue; however, i think going by our prior analysis the "share_ type_destroy" method here could use a db retry on deadlock. If multiple quota operations are occurring at the same exact instant, with no retry capability, we're open to hitting a deadlock - we have retry decorators on many other "update"/"delete" operations in the database code. We should probably add one and see if this helps.
I'm going to set this to "low-hanging-fruit" to see if any new contributors can pick this up.