ComplexIOd
commented
1 year ago Still, using 200% CPU right now (21:13)
While the status page shows that nothing is happening:
Open ComplexIOd opened 1 year ago
ComplexIOd
commented
1 year ago Still, using 200% CPU right now (21:13)
While the status page shows that nothing is happening:
nydave69
commented
11 months ago I have seen this problem (not in docker) In my case I had to split the number of show between instances of Medusa.. Database queries were taking toooooo long and backing up everything (status page usually show everything is fine)
I have also see this when loading the existing show page, it takes off scanning the folders without any interaction. As that's happening a banner on the bottom right will say loading x of y shows, and if x and y don't match well good luck. I have not found a good way out of this one other than to restart Medusa and that doesn't always solve this issue.
that's my 2 cents..
Aesculapius
commented
10 months ago what was the sweet spot for you when splitting up the instances? I now have around 2000 shows in it, should I half it?
Describe the bug While importing new existing shows from disk, the process is stalled by Medusa getting slower and slower. It is running in a docker through Portainer (latest) and nothing special is done triggering this. The database is somewhat large maybe (1700+ shows) but that shouldn't make the webpage unresponsive should it? After having done a few shows, the UI is immensely unresponsive making adding more shows impossible. It's not really a CPU spike, but a continuous blast of 100% CPU usage (see screenshots).
The docker is running on a virtualized machine which has 24 cores made available to Portainer. But Medusa is only using one of them it seems so one question would be in what manner Medusa is truly multi-threaded?
I'm checking CPU load by statistics in Portainer, running htop in the host machine and within the Medusa docker and by monitoring with Glances running in separate docker (but with access to other dockers).
To Reproduce Steps to reproduce the behavior:
Expected behavior For the GUI to stay responsive even when backlog is increasing
Screenshots
You can't really see the slowness on a screenshot... but all of the buttons become unresponsive...
CPU usage 100%, from Glances dashboard:
Stats from Portainer / Medusa docker container:
HTOP from host:
TOP from container:
Medusa (please complete the following information):
Debug logs (at least 50 lines): General > Advanced Settings > Enable debug
Not sure what part to paste...
Additional context Add any other context about the problem here.