Hi, thanks for great product.
Im trying to integrate crash reporting in our intranet, backed by sentry-native SDK (master 0.3, Windows+Crashpad) and on-premise sentry.
I used master branch as is, excluding nginx (modified to use external nginx) and pretty much copied all nginx settings.
I noticed that, crashpad handler working as expected generating minidumps, and sending to my sentry instance with HTTP/OK (verified with wireshark), in all my tested occurences.
But I noticed that if my minidump is less than mb size, everything is ok, but if my minidump is 2mb, I see
[relay_server::actors::events] ERROR: error processing event: could not store event
май 27 10:39:04 svn docker-compose[5375]: relay_1 | caused by: failed to send kafka message
май 27 10:39:04 svn docker-compose[5375]: relay_1 | caused by: Message production error: MessageSizeTooLarge (Broker: Message size too larg e)
(ignore datetime in russian, sorry)
and event is not delivered to sentry web UI.
I did not touch https://github.com/getsentry/onpremise/blob/83160e8bdbebfea2c0f02849b545b9cf1b30c324/docker-compose.yml#L86-L87 KAFKA_MESSAGE_MAX_BYTES and KAFKA_MAX_REQUEST_SIZE, so 50 mb should be fine in my case. It could be kafka not pickuping these settings, but I not kafka expert, so I’m not sure how to veriify this?
I can send you my modified docker-compose(basically just switched off nginx, and exposed ports to web and relay) if needed. Nginx access log is fine, and I’m pretty sure docker internal network case, but I could be incorrect.
Any suggestions what can be wrong with my setup?
Thanks,
Alex