Metabase docker container crashing in version 50.20 / 50.21

Bug reports
1

Hello,

I recently updated my Metabase to 50.20 and noticed it would crash randomly with a JVM OOM error.
I updated the memory allocated to JVM but it still crashed.

I also tried version 50.21 but there was no changes.

Were there any major changes in these version that would be causing this issue?

It seems like the container crashes after 12-14 hours.

2

Also, is there an equivalent of this command for docker?

java -Xmx2g -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/path/to/a/directory -jar metabase-jar

3

can you please send logs and what's the process doing when it crashes?

4

@dragonsahead Which logs? Also, how do I figure out the process when it crashes?

5

Metabase, as any other application, has logs. Those logs appear as soon as you run the application in interactive model (e.g. java -jar metabase.jar), or can be sent to some place else if you run it detached (e.g. log to a file). Please send the logs of Metabase so we can see what's going on

7

Hi @dragonsahead were you able to review the logs? I am still noticing these crashes occur with version 50.21.

8

The logs speak for themselves: the channel in slack no longer exists

9

(post deleted by author)

10

oh I didn't see the crash, Metabase is running out of heap space, so you need to increase the RAM in your server or at least increase the heap space that the JVM is allowed to use

11

Yeah, I did initially. Originally the JVM was allocated 1GB but I changed it to 3GB and it still crashed eventually.

12

please send how much RAM the server has and how much RAM does Metabase have (you should see that when Metabase starts)

13

8GB server and 4GB Metabase.

14

@dragonsahead I tried updating to version 50.23 on my instance noticed it would crash right away with the same JVM OOM error.

I tried allocating more RAM to the JVM -4GB and the host has 8GB but this didn't stop the crashes.

Any ideas?

15

Please assign 7GB to the application and let us know how it goes. Send also the logs when it crashes

16

@dragonsahead Sure, I can try that, but the logs are similar as the previously one I sent where it shows normal activity then suddenly crashed due to JVM OOM.

It definitely seems something is different as previously on version 50.15 the system never crashed and Metabase was only allocated 1GB out of the 4GB memory available.

17

Just FYI: we haven't had a single customer with memory crashes since we fixed the issues in the early versions of 50

18

Interesting.

Is there any other setting or parameter I can add to the docker to get more verbose logging to see what could be causing the crash?

19

nope, just send us the logs

20

(post deleted by author)

21

(post deleted by author)