Out of memory failure, but Metrics page doesn't agree

1

I just got an email notifying me that one of my services just failed. Consulting the dashboard I can see that it got OOM killed:

Screen Shot 2021-11-17 at 4.39.41 PM
Screen Shot 2021-11-17 at 4.39.41 PM760×468 55.7 KB

but on the Metrics page, I’m not seeing that service ever exceeding 400M:
Screen Shot 2021-11-17 at 4.40.21 PM
Screen Shot 2021-11-17 at 4.40.21 PM1920×724 35.6 KB

A quick perusal through the Logs doesn’t show me anything either.

Why was my service killed?

2

Hi @samuela, I took a look at your service and it was killed for being out of memory. The metrics you see on the dashboard are an average over two or three minutes so so a sudden spike won’t show on the graph reaching 100% of the limit.

3

Hi @samuela, I took a look at your service and it was killed for being out of memory. The metrics you see on the dashboard are an average over two or three minutes so so a sudden spike won’t show on the graph reaching 100% of the limit.

[Discourse post]

4

How can I see the max over 2-3 minute intervals instead of the average? The average doesn’t really help debug the issue since this is a case of spiking memory usage.

5

Hi @samuela, apologies for the delayed response. Right now our dashboard metrics are limited to that 2-3 minute average. Please feel free to submit that to feedback.render.com so we can help prioritize it and you can be notified of updates.