With that realisation, that it may be a bug in PMS may i kindly direct you to reread:
Older Versions of PMS did not do the Intro thingie, so no idea. Normal Transcode Operations go in the correct Directory, its just the Intro Detection that spills over into TMPDIR
BTW while you are talking to the dev Team, say thank you for the Skip Intro Feature, this is really great and implemented in a clever way
Thank you for tolerating me and my slowness.
In the 3 days this week so far, I’ve fixed 7 different bugs and introduced initial forum preview testing for the RPM packaging upgrade (so it’s like Debian)
Brace yourself 
Engineering already knows and is on it.
I just asked the transcoder team and the reply was that the head of the team who did skip intro knows about the regression.
We might get a hotfix for it
Temporary workaround? Move TMPDIR
Great news. Thank you.
Already done with Intro Detection on my Library though.
What I did was binding a hdd dir to /tmp
Intro Detection Done: 19332
Intro Detection Missing: 0
mount --bind
would do it too. GREAT thinking
Again, I am sorry as HECK for being dumber than box of rocks.
I did NOT connect the dots
This is NOT my norm.
I agree that an SBC is not the best reference case to validate everything against, however I bet that for very most server software, including PMS, the share of SBCs it is installed on is massively increasing, since at least the newer generation are greatly capable of doing nearly everything (great GPU acceleration support included, more and more even in mainline Linux) in small office or home environments. Not a killer argument but something that could be kept in mind 
.
The 512 MiB default for tmpfs, hence /tmp on the other hand is not related to the underlying board.
I was not sure if there is any practical difference between intro scan transcoding and “regular” playback tanscoding so that there is an argument to use different cache directories. In case, there surely was an argument to put the transcoding cache by default into /var/lib/ instead of TMPDIR, and the same argument would apply for intro scan transcoding as well?
Basically I vote for keeping TMPDIR untouched and use it as it is/was for small tmp/runtime files while, for simplicity, the already existing (+setting in web UI) separate TranscoderTempDirectory for both, playback transcoding and intro scan transcoding, following Orko.
I bet it is already implemented greatly but it came into my mind that for intro scanning of course only a short time of the video needs to be transcoded to catch the intro, however either this is already practice or technically not as easy as my limited insight makes me imagine 
.
Engineering is already working this and there are other threads about this issue.
Engineering admitted it’s their problem and is already being resolved.
I can see, that the scheduled tasks might (also) have something to do with the issue: The tasks are scheduled for 2:00 A.M. and at this time, I can see /tmp Filesystem starting to get filled up with files to 100% by some plex processes. And with 100% /tmp usage, my NAS stops behaving normal and needs to be restarted.
Settings - Server - Library - ‘Generate intro video markers’
set to “never” until this issue is fixed.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
