After 3 months of my app running normally now it shutted down due to lack of memory

Questions / Help
1

Yesterday day 7 my app stopped working, I check my fly.io account and there was a warning and I assumed the reason my machine stopped was because of the warning. (I don’t quite remembered what it was but it was probably due to the [Platform disruption] incident or something with the Madrid servers)

Today I woke up and the app was still not working, and I think it was stopped even tho I specifically had in the fly.toml to keep always one instance running (and for the last months my app never stopped automatically)
I went to my dashboard check my app logs and it looks like it’s always trying to start its machines and failing (the database server is working fine)
Then I received an email saying that my app crashed because It ran out of RAM even tho it worked fine for the past 3 months.
I tried to redeploy, it went through the process fine but my machines immediately went down and we were back to the same cycle.

I’m thinking about killing the machines and get new ones or increase the memory and then decrease it again and see if it works.

Anybody knows if this sudden request for more memory after working fine for 3 months is normal? Also the app didn’t originally went down because of memory but because of some incident, but know it shows that there are no status problems which is kinda weird.

Thanks in advance!

1 Like
2

Hi,

I’ve no idea about the cause but other users have reported similar issues e.g see this thread. If your app is set to 256MB, I wonder if it’s connected:

2 Likes
3

Yes, the pattern I’m seeing is that boots on 256MB machines will OOM where they previously booted and ran under load just fine. I’m working around it temporarily with the addition of swap, but I’m hopeful that we see a response from Fly on the topic and see resolution. Their responses on the thread I link to are quite discouraging so far though :confused:

1 Like
4

I agree, is the same pattern.
I get the out of memory error when deploying with a shared-cpu-1x 256

[info] [ 4.497482] Out of memory: Killed process 318 (beam.smp) total-vm:1785404kB, anon-rss:81124kB, file-rss:0kB, shmem-rss:73716kB, UID:65534 pgtables:472kB oom_score_adj:0

But when I upgrade to shared-cpu-1x 512 it works just fine, even tho the error before said that the total-vm needed was 1785.404MB.
It definitely seems to be a problem with the 256MB machines.

1 Like
5

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.