JS Push consumers crash the process when the NATS server goes away

[OtaK]

Observed behavior

When the NATS JS server goes away (connectivity issue, etc), the owning process gets crashed because there's an unwrap here:

nats.rs/async-nats/src/jetstream/consumer/push.rs

Line 503 in 7a094ff

context.client.state.changed().await.unwrap();

Expected behavior

Not crashing the owning process

Server and client version

latest for both

Host environment

N/A

Steps to reproduce

No response

[Jarema]

Thanks for the report.
Will issue a fix for it.

[Jarema]

Do you happen to have reproduction example?
I'm not sure how State Sender can be dropped (and cause a panic) if the whole client is not being dropped.

[OtaK]

My repro is using a NATS KV watch_all operation (which creates a push consumer under the hood) and having the NATS server be SIGKILL'd, this creates this panic.

At the same time even if the client is dropped as a whole it shouldn't panic its owning thread either way, because this prevents graceful recovery