Error setting/closing connection: Not Connected

Hi everyone,

I’m fairly new to Metabase, and while I’m mostly loving it so far, I am having an issue that I haven’t yet been able to get to the bottom of.

When I scroll through the logs, I am frequently seeing errors like the following:

May 08 09:00:00 ERROR metabase.pulse.render :: Pulse card render error clojure.lang.ExceptionInfo Card has errors: [Amazon](500150) Error setting/closing connection: Not Connected. 

As referenced above, it happens commonly (though not exclusively) during pulses that run at night. I tried searching on that error code and/or error message to no avail. Looks like the connection is dying, maybe timing out? Has anyone seen this before. Are there connection pool settings that I can tweak to get more reliability?

I’m running 0.32.4 and the database in question is Amazon Redshift.

Thanks in advance!

Hi @JZee
Could you try updating to 0.32.5
It looks like the database connection gets lost/dropped. Do you see anything elsewhere in the log that could hint at that? Or could it be that Metabase has used all 15 connections to Redshift, so there isn’t a new connection available?
I’m not sure if Redshift provides logs, but could you check around the same time as you’re seeing Pulse error?
There’s a handful of issues related to “Pulse card render error”, but I don’t think you’re seeing one of those.

Hi @flamber,

Yup. I realized that I’m one version behind about 10 seconds after I sent that post :slight_smile:. Upgrading now. I’ll let everyone know if that solves it.

I’ll keep digging through the logs for clues, but so far it’s nothing obvious. The thought occurred to me that I might be running low on connections, but that would be surprising, as this thing is not under much load.

Thanks for the ideas!