Monitoring best practice

[Hechen Gao]

Hi Jaeger community,

I am planning to integrate jaeger into our system for tracing, and I am working on deployment process right now.

One question I have is regarding the monitoring of jaeger components. What are the things you recommend i should monitor for jaeger-agent, jaeger-collector and jaeger-query?

[Yuri Shkuro]

we monitor

1. the usual process signals like CPU/memory usage, file descriptors, network usage
2. Go runtime signals like goroutine count, GC
   
3. Jaeger internals: dropped span count, mem queue size, storage write counts/latencies

--
You received this message because you are subscribed to the Google Groups "Jaeger Tracing" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jaeger-tracing+unsubscribe@googlegroups.com.
To post to this group, send email to jaeger-tracing@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jaeger-tracing/3873093e-26a8-4e0a-8eac-5bb23d0f5f81%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Hechen Gao]

Thank you so much for the response Yuri,

That's quite helpful. I am also planning to deploy jaeger-collector and jaeger-query as Kubernetes HPA, so that it can auto scale on certain conditions.

Do you happen to have similar auto-scaling setup? If so, what are the conditions you use for auto-scaling.

Best regards,

Hechen Gao

On Tuesday, February 6, 2018 at 4:57:08 PM UTC-8, Yuri Shkuro wrote:

we monitor

1. the usual process signals like CPU/memory usage, file descriptors, network usage
2. Go runtime signals like goroutine count, GC
   
3. Jaeger internals: dropped span count, mem queue size, storage write counts/latencies

On Tue, Feb 6, 2018 at 7:52 PM, Hechen Gao <zere...@gmail.com> wrote:

Hi Jaeger community,

I am planning to integrate jaeger into our system for tracing, and I am working on deployment process right now.

One question I have is regarding the monitoring of jaeger components. What are the things you recommend i should monitor for jaeger-agent, jaeger-collector and jaeger-query?

--
You received this message because you are subscribed to the Google Groups "Jaeger Tracing" group.

To unsubscribe from this group and stop receiving emails from it, send an email to jaeger-tracin...@googlegroups.com.
To post to this group, send email to jaeger-...@googlegroups.com.

[Yuri Shkuro]

Hi,

No, we don't use auto-scaling for collectors, just a fixed size pool. Even if we did, the bottleneck is typically the storage, which is hard to auto-scale.

To unsubscribe from this group and stop receiving emails from it, send an email to jaeger-tracing+unsubscribe@googlegroups.com.
To post to this group, send email to jaeger-tracing@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jaeger-tracing/56f86d81-ea59-432b-8735-cbcb6b9524da%40googlegroups.com.

[Hechen Gao]

I see, thank you so much Yuri for the guideline, this is super helpful.

Best regards,

Hechen Gao

[die...@grafana.com]

Are there any prometheus dashboards for jaeger monitoring, to import in a grafana instance?

a good practice is to maintain a dashboard in the code repo and publish it on https://grafana.com/dashboards

thanks,

Dieter

[Won Jun Jang]

It has been discussed: https://github.com/jaegertracing/jaeger/issues/789

To unsubscribe from this group and stop receiving emails from it, send an email to jaeger-tracing+unsubscribe@googlegroups.com.
To post to this group, send email to jaeger-tracing@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jaeger-tracing/9ed7c821-0c8f-4fcd-b84f-722529e69975%40googlegroups.com.