Exporter: timeout leak

[bernardo-martinez]

Hey, I am running opentelemetry_exporter 1.8.0 in elixir project and I get a loop of timeouts after first timeout of my collector not being available.

The thing is that when my collector (grafana alloy) restarts I see a log of:
OTLP grpc export failed with error: :timeout","level":"info","mfa":"{:otel_exporter_otlp, :export_grpc, 5}"}

And then it keeps looping on that every 5seconds until eventually the services gets OOM (memory goes from 5oomb to 1gb in aprox 40minutes) despite the collector being available

I can provide further details if needed, thanks!

[tsloughter]

Thanks! I'm guessing the IP changes of alloy? My best guess is there is some issue handling the fact that one of the IPs in the channel is no longer valid. Surprised this only comes up now for the grpc library if that is the case, but seems the most likely if it happens on every restart (assuming it gets a new IP or port)

You can switch to using HTTP for now which shouldn't have the issue while I debug this.