Hey, I've noticed a very weird behavior in Pulsar. I'm emitting events to a non-persistent topic and use a Shared consumer on it. For some reason, if I look at the stats of the topic, I will see unacked messages even though they were acked. Here is a minimal example: https://gist.github.com/psisoyev/87db4f11ec28125e1af3988d402af41c Now I wonder, what could be consequences of this? I found it while investigating weird behavior of my producers, at some point they couldn't connect to Pulsar:

│ [pulsar-client-io-1-1] WARN  o.a.p.c.i.ClientCnx - [id: 0x43fa742b, L:/***:34102 - R:pulsar-prod-resources-broker-1.pulsar-prod-reso │
│ urces-broker.pulsar-prod.svc.cluster.local/***:6650] request timeout {'durationMs': '7000', 'reqId':'58081229739110752', 'remote':'pulsar-pr │
│ od-resources-broker-1.pulsar-prod-resources-broker.pulsar-prod.svc.cluster.local/***:6650', 'local':'/***:34102'}

even though all the network is working fine, this just happens from time to time and I have to restart my brokers and then it works well. Im using 2.11.2 Pulsar. Any advices on what can I do to mitigate the issue?