Metabase Timeouts
Metabase 1 provides a nice way to retrive business intelligence data from your databases, quickly and beautifully.
Expectedly, most users fall down the hole of executing heavier and heavier queries, and crash against the wall that is this message: “Your question took too long”. Timeouts.
Most advanced users will quickly realize that their proxy will need some special timeout header, or a value increment of it. However, there is another parameter hiding out there, that I happened to find recently, and relates to Jetty2.
Setting this environmental variable MB_JETTY_ASYNC_RESPONSE_TIMEOUT and a value in miliseconds will increase what the name describes, and will save you the pain of trying to find out why even though you set your proxy timeouts at 20,30,60 minutes, your queries still die after 10 minutes. For example a value of 900000 means 60s * 1000(milis) * 15, so 15 minutes.
Then, if you would like to apply this using the official (now deprecated) Helm chart, you could add these lines:
jetty:
ASYNC_RESPONSE_TIMEOUT: "900000"
Even though the values.yaml file does not mention this property, if you take a look at the template file you’ll understand why this works.
Hope I saved you some time, in trying to find out why your Metabase queries timeout.