Metabase shows blank page after some time

Get Help
1

Hi, I have been running Metabase in production for a business use case in AWS Fargate. The Container spec I have given for it is 4vcpu, 18GB RAM and I am running a single container. It is exposed via ALB

Now, when I run it, it runs fine for a few days, and after about 10 odd days, the Metabase hangs up, won't show the SSO login screen, and shows a blank white page. What could be the original fix for it? Is it Java opts not rendering inside properly?

Current Java Opts :

                {
                    "name": "MB_JETTY_ASYNC_RESPONSE_TIMEOUT",
                    "value": "1500000"
         
                    "name": "MB_JETTY_MAXTHREADS",
                    "value": "800"
                },
       {
                    "name": "JAVA_OPTS",
                    "value": "-Xmx18g"
                },
                    "name": "MB_JDBC_DATA_WAREHOUSE_MAX_CONNECTION_POOL_SIZE",
                    "value": "300"
                },
2

That's a massive instance, Metabase only needs 1GB per 20 concurrent users...

also your Java_opts env var is assigning 18 gigs to the process but the container also runs other processes! and also you're not leaving memory for the non-heap

please post the version

reards,
Luis

3

Metabase version : 0.50.28

I have updated the JAVA OPTS to 16GB and am still keeping 18GB in the task though. got some headroom

Modified the following too :

  • MB_ASYNC_QUERY_THREAD_POOL_SIZE : 100
  • MB_JETTY_ASYNC_RESPONSE_TIMEOUT : 2500000
  • MB_JETTY_MAXTHREADS : 1200
  • MB_JDBC_DATA_WAREHOUSE_MAX_CONNECTION_POOL_SIZE : 600
  • MB_APPLICATION_DB_MAX_CONNECTION_POOL_SIZE : 100

Added this as well to the task def :
"ulimits": [
{
"name": "nofile",
"softLimit": 1048576,
"hardLimit": 1048576
}
],

let me know if I am doing something wrong