[🐛 BUG]: panic in centrifuge plugin used by buggregator

[devnev]

No duplicates 🥲.

• I have searched for a similar issue in our bug tracker and didn't find any solutions.

What happened?

Although this is a failure of a buggregator instance, it looks like its entirely in roadrunner, so I'm filing it here.

I'm running a bunch of services with a docker compose, including a spiral application, buggregator, jaeger, nginx, etc.

I've been encountering startup issues with buggregator, and I've now even gotten a panic.

The panic is:

buggregator-1   | panic: runtime error: invalid memory address or nil pointer dereference
buggregator-1   | [signal SIGSEGV: segmentation violation code=0x1 addr=0x28 pc=0xc61bb2]
buggregator-1   |
buggregator-1   | goroutine 228 [running]:
buggregator-1   | github.com/roadrunner-server/centrifuge/v4.(*rpc).Broadcast(0xc00097db30, 0xc0008d0018?, 0xc000db0f80)
buggregator-1   | 	github.com/roadrunner-server/centrifuge/[email protected]/rpc.go:81 +0x52
buggregator-1   | reflect.Value.call({0xc0000c3440?, 0xc0001adf20?, 0x13?}, {0x1f315f7, 0x4}, {0xc000014ef8, 0x3, 0x3?})
buggregator-1   | 	reflect/value.go:596 +0xce7
buggregator-1   | reflect.Value.Call({0xc0000c3440?, 0xc0001adf20?, 0xc0009c02b0?}, {0xc000a4c6f8?, 0x411845?, 0x10?})
buggregator-1   | 	reflect/value.go:380 +0xb9
buggregator-1   | net/rpc.(*service).call(0xc0000509c0, 0xc0009c0270?, 0xc000db0280?, 0xc000da81a0, 0xc000793a00, 0xc000a4c7d0?, {0x1e015e0?, 0xc0007e7180?, 0x24b7cc0?}, {0x1d96be0, ...}, ...)
buggregator-1   | 	net/rpc/server.go:382 +0x214
buggregator-1   | created by net/rpc.(*Server).ServeCodec in goroutine 227
buggregator-1   | 	net/rpc/server.go:479 +0x410
buggregator-1 exited with code 2

Before this there were repeated logs of

buggregator-1   | 2024-11-13T14:20:02+0000	ERROR	service.centrifuge	wait	{"error": "exit status 1"}

Version (rr --version)

The buggregator docker image is

ghcr.io/buggregator/server:1.11.3

The dockerfile at that commit has ARG ROAD_RUNNER_IMAGE=2023.3.7 so I presume the version of roadrunner is

2023.3.7

How to reproduce the issue?

I can reproduce the ERROR service.centrifuge wait {"error": "exit status 1"} when I simulate heavy CPU contention (as happens during the start of services with docker compose) with the --cpus flag, e.g. docker run --cpus=0.1 ghcr.io/buggregator/server:1.11.3

So far I have not been able to reproduce the panic.

Relevant log output

No response

[rustatian]

Hey @devnev 👋
Could you please update to v2024.2, because v2023.3.7 is hugely outdated.

[devnev]

@rustatian OK, I've raised an upstream PR: buggregator/server#261

[rustatian]

Have you tried the latest version? Are there any problems with it?