Go routine leak in azure-event-hubs-go/storage.(*LeaserCheckpointer).GetLeases

[jonnylangefeld]

Expected Behavior

Amount of go routines for azure-event-hubs-go/storage.(*LeaserCheckpointer).GetLeases.func1 doesn't increase over time

Actual Behavior

the go routines are leaking and memory footprint grows

Analysis

I found the same issue as described in #92 in the function azure-event-hubs-go/storage.(*LeaserCheckpointer).GetLeases

I have a piece of software, that is reading from eventhubs and writes simple events to a mongodb.
The following two screenshots of flame graphs of pprof were just taken 5 minutes apart. You can see the growth of the azure-event-hubs-go/storage.(*LeaserCheckpointer).GetLeases.func1 go routine, while all others stay the same:

Running the software for 5 days (with millions of events daily) leaves us with growing go routines like the following, as long as the pod is not restarted:

Proposed solution

The commit that fixes #92 introduced the defer close(resCh).
This probably has to be added for leaseCh after line 193 in azure-event-hubs-go/storage.go as well.

Environment

• OS: linux
• Go version: 12.9
• Version of Library: 2.0.0