We are using an on-premises open-source Metabase project. We deployed Metabase on an EC2 instance in the Singapore region, which connects to a MySQL RDS in the same region. We run the Metabase jar file using the command "java -jar metabase.jar". We launched this setup on December 17th, 2022.
However, within six days, the AWS RDS server shut down and displayed a "storage-full" error. To resolve this issue, we increased the storage capacity to 1000 GB on December 25th, 2022. Unfortunately, we encountered the same "storage-full" error from RDS on January 1st, 2023, and subsequently. We have now shut down the Metabase server as it continued to consume RDS storage. We are still unable to identify what kind of data Metabase is storing in our RDS server.
Please note that we have not migrated from an H2 database.
can you get inside the rds and see if it has really consumed the 1GB of space, which is completely weird as the metabase metadata won't consume that in a 100 years unless you have a massive quantity of users?
Actually I have shut down the RDS server as we saw this weird behavior. Even i found this very unusual,
Suppose if we point Metabase to Replica RDS , Then this situation will not occur, as Replica by default write operation is blocked, So even metabase tries to write on Replica, It wont be possible & i wont face any storage full issue.
If Metabase can't write to the app db, then your users won't be able to authenticate, questions won't be able to be saved, etc etc. There's something going on there which you need to investigate about why your RDS is skyrocketing.
you need to move to a postgres or mysql as the application database. If you don't do that, please consider seriously that you will lose all your work sooner or later due to corruption of the H2 embedded database
if you run out of size in the data warehouse you're using, then it's not Metabase fault then, it's something else. Metabase only reads from the DW, it never writes unless you have enabled model caching