[Bug] Crash report in #4423 doesn't seem to be resolved

[cylewitruk]

Please see closed+merged ticket #4423 for more information.

Tagging @8marz8 and @badonyx

[8marz8]

@badonyx Can you confirm which branch/instance you're using? Because the code you referenced was already changed to introduce some retry logic:

https://github.com/stacks-network/stacks-core/blob/next/testnet/stacks-node/src/config.rs#L2036

[badonyx]

I am on an older branch. It looks like it might be resolved with that change. I will try it out.

[badonyx]

The problem with this fix is that it still panics and kills the node if there is an outage lasting longer than ~60 seconds.

The connection should be retried indefinitely, at least for a follower node; for a miner node maybe this should be configurable.

[cylewitruk]

The problem with this fix is that it still panics and kills the node if there is an outage lasting longer than ~60 seconds.

The connection should be retried indefinitely, at least for a follower node; for a miner node maybe this should be configurable.

I think it should be configurable primarily via environment variable, how many seconds to wait (0 = indefinite). Default should be indefinite, imo.

Environment variable makes it very easy to configure in k8s etc. where containers dying are a normal part of orchestration and monitoring, and I don't really see the use-case for non-indefinite via config file for other scenarios.