sorry, I need you to send us the logs before the thread dump starts, I need to know what Metabase is doing before it runs out of memory
this is extremely interesting. What data types do you have in your snowflake and mysql? can you separate both sync/scans?
There's a variety of data types integer, strings, Boolean etc. and majority of the database scans I have no run at 12 AM.
You want me to split the Snowflake / MySQL to different times?
After observing this more, it definitely seems like something is causing a memory leak that's leading to the crash. On this test instance I have, its idle and no one uses it and the only thing running would be the standard database syncs, etc.
And depending on how much memory I allocate to the JVM, it changes the duration of the crash. More memory = takes a bit longer but eventually crashes.
FYI- I updated my test instance to 50.24 and so far no JVM crashes and the container has been running for 38 hours so far. I also reverted the custom JVM memory allocation as well so now its using the default allocation parameter.