I’ve been successfully running Metabase for the past 10 months. About a week ago the pulse emails stopped going out. It’s worth noting, when performing queries in metabase manually, things seemed to be working.
When explicitly pressing the Test Email, the email would go out, but it would not go out at the scheduled time.
After deleting new pulses, and trying to pinpoint what changes happened in our database, we decided to simply restart metabase:
sudo docker restart metabase
This did the trick and pulses were delivered after the fact. If this ever happens to anyone in the future, give restarting metabase a shot.
Huh. That’s … not good.
Do you have the logs from the previous container lying around? There should be an error message in there that might narrow down why that happened.
I did review the logs. I did see a couple error messages about bogus SQL queries trying to execute - but it’s worth noting, nothing was coming up in the logs when pulse was supposed to be scheduled (prior to restarting it, I ran _sudo docker logs --tail 100 -f metabase_and was watching the logs when the email was supposed to go out and it didn’t), .
My hunch was the bogus SQL queries were caused by someone else manually running a query.
I didn’t have any logs from right when the emails stopped going out a week ago, so there’s a chance the server temporarily ran out of memory or something. The app itself functioned flawlessly though, only the pulse emails did not go out.
Prior to that the server had been up for 10 full months. I’m also on an older version of metabase, I’m planning to upgrade to the latest version soon.
Honestly, restarting the server doesn’t bother me. I did some Googling about the issue though, and I didn’t find any information for similar problems for other people, so I figured I’d report this to potentially help someone else in the future.
By the metabase is a terrific product. I’ve been very happy with it and it’s been remarkably helpful!
Hey did you find the solution to this ?
I am facing this issue since months now
See this topic and GitHub issue mentioned there for some workarounds and different reasons it can happen: Pulse report not going
Other than that, what’s needed to resolve issues like this one is details on your config /steps to reproduce. It realize steps to reproduce are particularly hard in a case like this — nonetheless it’s important.
I believe work is ongoing to add more debugging/logging info in this area, but it will take a while before there will be effects from that.