what s a canonical way folks have got their lambda logs into SigNoz Community #general

Join Slack

what's a canonical way folks have got their lambda...

# general

Ishwor Gurung

11/30/2022, 5:11 AM

what's a canonical way folks have got their lambda logs into signoz? also, what's recommendation? thankyou

Pranay

11/30/2022, 5:51 AM

@nitya-signoz should have more insights on this

Ishwor Gurung

11/30/2022, 5:54 AM

thanks Pranay

Ishwor Gurung

11/30/2022, 5:56 AM

i've got zapped log fields all ready to go from lambda with otelcol-contrib collector layer with signoz collector on k8s ready to rendezvous from lambda subnet (AWS)

nitya-signoz

11/30/2022, 8:14 AM

Don’t have suggestions as of now, but from what I understand from the above statements you have already got it working right? You just want to know what can be other ways to achieve the same/best practices?

Ishwor Gurung

11/30/2022, 8:16 AM

well, essentially yes. imagine greenfield lambda loggin to its stdout/stderr as per usual and now need to land those logs to signoz otel log collector. what's the best practices? there are tons of good documentation on non-lambda in the signoz.io website but couldn't find much on lambda based deployments 🙂

Srikanth Chekuri

11/30/2022, 8:35 AM

AWS maintains a lambda extension to make working with opentelemetry easy https://github.com/open-telemetry/opentelemetry-lambda. I didn’t try it myself, but I have seen people use it to send traces/metrics/logs to another vendor. You may want to give it a look.

Ishwor Gurung

11/30/2022, 8:59 AM

thanks

Ishwor Gurung

12/01/2022, 6:10 AM

So, a current interim solution that Nitya pointed out a week ago (thanks Nitya) is to use

filelog

receiver (we built custom lambda extension based on

opentelemetry-lambda

with

filelog

receiver). that receiver then hands off log (with traces as zap fields) to otlp exporter. lambda->/tmp/app.log->filelogreceiver->otlpexporter->signoz. this has worked so far.

Ishwor Gurung

12/01/2022, 6:11 AM

until Go client library+sdk supports logging signal out of the box, this interim solution works for us.

Ishwor Gurung

12/01/2022, 6:13 AM

/tmp

is recommended in lambda because AWS says it can scale from 512Mb to 10G.

Ishwor Gurung

12/01/2022, 6:13 AM

(enough for us)

nitya-signoz

12/01/2022, 6:55 AM

Interesting, thanks for sharing this. @Ashu I think we can create a blog/tutorial with inputs from Ishwor on collecting AWS lambda logs.

Ashu

12/01/2022, 6:57 AM

Thanks @nitya-signoz , that would be great. Let me reach out to @Ishwor Gurung with some plans for that.

102 Views

Open in Slack

Previous Next