Troubleshooting

Fixes for the problems you are most likely to hit with the OpenTelemetry Collector Gateway.

A client cannot connect to port 4317 or 4318

Cause: the client uses plain HTTP or does not trust the instance's certificate.

Fix: use TLS and trust /usr/local/share/sushi/otelcol/cert.pem. For OTLP HTTP, use https://127.0.0.1:4318. Plain HTTP is refused by design.

The Collector does not start after a config change

Cause: an error in /etc/otelcol/config.yaml, or a component that is not in this build.

Fix: read the error, check the component list, fix the file and restart:

sudo journalctl -u otelcol -n 50 --no-pager
/opt/otelcol/bin/otelcol components
sudo systemctl restart otelcol

Contrib-only components, such as the AWS exporters, cannot be added to this build.

Data is dropped under load

Cause: the memory limiter refuses data when the Collector nears its 512 MiB limit.

Fix: reduce the incoming rate or batch size at the source, or raise limit_mib in config.yaml if the instance has memory to spare.

Clients fail after restoring a backup

Cause: the public copy of the certificate no longer matches the restored key.

Fix: run sudo systemctl restart sushi-otelcol-firstboot, then restart the Collector.

First boot failed

Cause: sushi-otelcol-firstboot.service failed, so the Collector did not start.

Fix: look at its log, then run it again and start the Collector:

sudo journalctl -u sushi-otelcol-firstboot.service --no-pager
sudo systemctl restart sushi-otelcol-firstboot
sudo systemctl start otelcol

If it still fails, email info@softwaresushi.com with the instance ID and the log output.