ConnWriteTimeout doesn't appear to work.

[justinreddit]

We've specified ConnWriteTimeout, however, we observe client side latencies exceeding the timeout and not appearing to timeout. Are there client side metrics we can use to see whether this is getting triggered?

[rueian]

Hi @justinreddit, please use context deadline instead https://github.com/redis/rueidis?tab=readme-ov-file#context-cancellation. For currently available metrics, please refer to https://github.com/redis/rueidis/tree/main/rueidisotel.

[liuzhaohui]

We've specified ConnWriteTimeout, however, we observe client side latencies exceeding the timeout and not appearing to timeout. Are there client side metrics we can use to see whether this is getting triggered?

I encountered a similar issue under the senario of setting a context deadline for a request but occasionly it would take much more time until it realy returned. It's the issue of the connection dial didn't respect the context which fixed in recently release v1.0.56. May be you can try to use context deadline as mentioned above by rueian.

[justinreddit]

@rueian any guidance on adding request latencies when using DoMulti? It'll be useful to see what client side latencies are per command.

[rueian]

To see the latencies of commands, you can use rueidisotel to collect OpenTelemetry Traces. There will be latencies attached to each trace. If you want more customizations please refer to https://github.com/redis/rueidis/tree/main/rueidishook.