Stop pool connection on SIGINT

[alexvanin]

If pool is configured incorrectly, application freezes at NeoFS connection until timeout. Context is not canceled on SIGINT (^C).
I think this more of a SDK issue, but we can investigate it from here.

[KirillovDenis]

Could you clarify the pool is configured incorrectly?

[alexvanin]

Try to start s3 gateway with s01.neofs.devenv:8080 endpoint without starting neofs-dev-env.
On my machine, gateway freezes and doesn't react on ^C until timeout.

[KirillovDenis]

It seems here we don't accept any context

[alexvanin]

Can you try to do a PoC with having application context in Dial?
All you need to do is to pass app context into rpc.Balance() as a client parameter: client.WithContext(ctx)?

[alexvanin]

We found that app sleeps for 20 seconds. Not clear where this timeout is set.

[KirillovDenis]

We found that app sleeps for 20 seconds. Not clear where this timeout is set.

It's interesting that when we start app via IDE (goland) the app sleeps only for 3 seconds

[KirillovDenis]

Can you try to do a PoC with having application context in Dial? All you need to do is to pass app context into rpc.Balance() as a client parameter: client.WithContext(ctx)?

Yes. This solve the problem (^C stops app now):

2022-09-23T12:12:50.272+0300    warn    s3-gw/app_settings.go:144       skip, empty address
^C2022-09-23T12:12:51.230+0300  fatal   s3-gw/app.go:134        failed to dial connection pool  {"error": "at least one node must be healthy"}
main.newApp
        github.com/nspcc-dev/neofs-s3-gw/cmd/s3-gw/app.go:134
main.main
        github.com/nspcc-dev/neofs-s3-gw/cmd/s3-gw/main.go:63
runtime.main
        runtime/proc.go:250

[alexvanin]

Blocked by nspcc-dev/neofs-sdk-go#343

[alexvanin]

Unblocked. I guess it will be good to update all our components that use SDK to avoid connection freezing.