Accepted events not showing up in project

On-Premise
1

Hello,

I’m new to Sentry and after hours of digging through documentations and topics I still couldn’t find an answer.
So, we have an on-premise instance (9.0.0 of version) and I try to setup Sentry into our project. I started with a simple Node.js application and set it up based on the guide just to check how it works.
Now my problem is, that under the Stats I can see that the events arrived but only a couple displaying under the project and I can’t even find the events by their id. Also there is a significant delay too, when the event sent I immediately see under stats that the number is increased but the few event which show up under the project appearing after 1-2 hours.

Can someone able to shed some lights for this weird anonamyl? Thanks!

2

My best guess would be an issue with your cron or worker instances. Can you verify that they work reliably and with enough resources? It is recommended to have multiple worker processes if your load is high (not necessarily from this app but in overall Sentry usage)

3

@BYK Hello, I am having the same issue. I run 2 worker hosts, CPU hovering around 20-30%. There is not much in cronjob log.

Anything else I should look for?

4

@trungdq88 - not really. If you provide more details about your setup, we may be able to find other places to look at though.

5

@BYK Thanks for getting back, here is my setup:
2 hosts

  • Host spec: GNU/Linux - 2 CPU - 4 vCPU - 10.213.253.12 - 7.84G - 110.25G
  • CPU usage: ~20%
  • Memory usage: ~5GB out of total 8GB
  • Each host run 3 master celery process x 4 worker processes each (12 worker processes in total)
  • client is using raven-python 5.21)
  • Redis CPU usage ~20%
  • Postgres CPU usage ~ 20%, number of connections is ~10
  • We have about ~700 events per minutes to our sentry (stats from sentry dashboard)

What else do you need? Thanks for helps!

6

Hi @trungdq88! Our resident scalability experts say it is most likely a job contention between fast jobs and slow jobs and the best remedy is focusing worker processes on specific queues. Getting some metrics and graphs on the queues it’d help making a more informed decision.

At this point I also cannot help but recommend at least trying the SaaS version on sentry.io if this is becoming hard to maintain :smiley:

7

Hi @BYK, thanks for getting back.

Getting some metrics and graphs on the queues

Where can I get these information? Are these available the internal Sentry metrics stated in Sentry’s document?

8

Sorry for the very late response. This may be helpful: https://docs.sentry.io/server/internal-metrics/

That said if you haven’t set this up earlier, I doubt it would be useful to debug your current issue :frowning: