I don’t know if it is or is not an issue.
I have it on now and creating a new TV-Intro library. Memory usage is still normal.
I’m running out of things to try.
Maybe turn off one or two of the other apps which run? (thinking that you might be just tripping a threshold which causes the avalanche )
Tonight I’m going to try to change the scheduled task hours in case there’s a conflict with the NAS. If it stays the same, I’ll try temporarily disabling 1 app (although it’s not the solution). If it stays the same, I’ll remove the intro detection. Yes
If even after all that it stays the same… Then we’ll have to find another solution.
I’ll keep you posted.
Thanks @ChuckPa
Look my logs (Plex Media Server.log.1) at:
Nov 22, 2020 07:39:29.642 [0x7f70146ab700] DEBUG - EPG[cloud]: Purging 0 airings which completed in the past.
Multiple
Nov 22, 2020 07:46:02.325 [0x7f7014f75700] WARN - HTTP error requesting GET http://192.168.1.53:4789/upnpd/49da96086c.xml (7, Couldn't connect to server) (Failed to connect to 192.168.1.53 port 4789: Connection refused)
(File removed)
I have disabled intro detection.
I have scheduled the automatic tasks for an hour from now.
I will be aware of what happens at that time.
It seems that it ended without major complications.
I have rerun scheduled tasks with intro detection turned on again. I’m going to see what happens in this case…
@ChuckPa.
With intro detection active the system has broken again.
Could it be a generic bug or can you tell if I have a corrupt video file?
It looks like a loop.
The htop output you show is “Thread View”. PMS has approx 75 threads within the process based on what it’s actually doing.
ps only shows processes.htop has the ability to show each individual thread. Turn off Thread view for best view of reality.RES size is the same and is only 137m (MB) for PMS.I can’t analyze from screenshots but a damaged file is a valid theory.
I will need the ZIP of the logs (captured right after it reports the activity) to see what is happening)
Hi @ChuckPa
Before restarted
htop
ps -a
4U (just restarted)
(File removed)
Plex Media Server.1.log
Nov 22, 2020 18:00:41.761 [0x7fbeebbe1700] DEBUG - Activity: registered new activity e55ab16b-ec6a-45e7-8750-e21348270d85 - Detecting intros
See how dangerous htop alone can be? It gives the wrong impression of what’s happening.
ps -ef and regular top are much better for this type of work. htop is too finely detailed when you’re not looking for something specific.
As a comparison, here is top, in color mode, on my workstation.
Red = sleeping and Green = active (processes)
I see a great deal of Music scanning.
Are you downloading music (or other media) directly into the folders PMS scans?
If so, that would explain all the activity I see.
Thanks for the htop /top/ps hint.
Regarding music … I’m not downloading anything. The music library is in a folder called “Música” within the media library. I have a music library of 3028 tracks.
This is a problem?
It appears to be scanning … and scanning… and scanning.
Is everything named properly (per plex spec) ?
I would like to suggest letting the machine settle down.
I would also like to suggest increasing the log file retention so we can get a better view of what is happening over a wider time interval.
To do this:
/> , add LogNumFiles="10"SomethingElse="Value" LogNumFiles="10" />)I just rebooted with that parameter active.
I have done an analysis of my music and it seems well renamed, at least Plex has recognized it thanks to the MP3 tags.
In my files when the system crashed … I saw this line …
Nov 22, 2020 18:00:41.761 [0x7fbeebbe1700] DEBUG - IntroDetector: Running intro detection for [39080] [Sobrenatural] [15]
Nov 22, 2020 18:00:41.761 [0x7fbeebbe1700] DEBUG - Activity: registered new activity e55ab16b-ec6a-45e7-8750-e21348270d85 - Detecting intros
It may be an episode of the “Supernatural” series that is causing the problem.
I can delete my music library if you want but I don’t think that’s the reason.
Intro detection is only performed for Television series episodes.
Effectively.
So I think it is a corrupt episode or a problem in the intro detection system.
My money is on the existence of a bad episode.
If intro detection were at fault, everyone would have problems such as this which isn’t the case.
Is it possible it could be more robust? Sure, first we need find out what the problem is.
When you are working through the video episodes,
When you think you have it (by trial & error exchange)
– May I please have a copy of the one which does cause you a problem?
Having a known problem video would let me give it to Engineering so they can improve how PMS behaves in those situations.
Yesterday I started playing my next unseen episode of Supernatural.
My surprise was when in that episode there began to be cuts and jumps in the video.
I looked for another version of the correct file, downloaded it and played it without problem.
I will run the scheduled tasks again in 20 minutes. Perhaps this badly played video file was causing the problem.
However, I am sorry to say that I replaced this file yesterday and that I no longer have the supposedly wrong video file … Sorry
It seems that in the last execution everything went well. It seems that it was related to the episodes of the series “Supernatural”.
I will be waiting the next few days to see how it evolves.
