See my post about video preview thumbnailing above
We need to leave the 3gb database / 1 million music tracks server out of this for now
Oh, I am. In fact, because of the oddities about that, Iām probably going to set up the new server with 1.14, so it wonāt be up for discussion. I doubt moving Audiobooks/Radio will make much difference to the size of the database, either ā Iām taking the advice about installing an SSD, too.
Iām considering whether or not I should install 1.15.1.725 on the Mouse. The main consideration is that the server wonāt be up all the time, so itās likely not useful to have that build.
All I said was I do not want that to be part of what we are trying to solve here - concentrate on the other server and let me know the outcome after you disable video preview thumbnailing
Relax, relaxā¦on the same page here. I understand the issue with the gigantic database, and Iām not concerned about leaving it out, nor am I concerned about leaving the Mouse out of the loop when thatās set up. Iām moving the audiobooks and radio libraries out of practicality more than anything; I donāt generally access either of those on a daily basis, so that server will mainly be offline.
Additionally, BearcubII will be going away and Iām going to be starting over on BearcubIII ā and Iāll just run that on the current public release.
That leaves just Brainbuntu, and the only logs and data youāll see coming from me will be from that installation.
Video previewing is disabled as of an hour ago. What do you want me to do with library tasks? Are there any settings I should change that might result in useful information?
Automatic library scan is OFF
Partial scan when changes detected is OFF
Scan my library periodically is ON, Daily
Empty Trash is ON
Allow media deletion is ON
Include season premiers in On Deck is ON
Run scanner tasks at a lower priority is ON
Generate video previews and chapter thumbnails are OFF
Audio analysis is OFF
And I think I just had one of those weird blip crashes, so Iāll be back with logs in a few minutes.
How odd. Nothing in Event Viewer, nothing in the CrashDumps folder. I actually saw it happen, though ā rollerblinded up, rollerblinded back down, about two seconds flat.
Plex Media (13.3 MB)
Have been looking at the logs. cannot see any crash. It may have been a transient period when the server took a long time to respond to specific requests to Plex Web - I noticed at 10:37 am requests took close to 20 seconds to complete instead of the usual 3-5 seconds max - for loading the recently added photos and music hubs to the dashboard
Was this the time ? (10:37 am - 6th March)
The lengthy response time coincided with the start of a scan of 37,000+ photos library.
Did you get failure to load dashboard error ?
Not at that time, no. The entire web UI acted as though PMS had crashed, with all of the libraries on the left going away and the dashboard going away ā just as though it had crashed or been shut down.
It just as suddenly came back, dashboard and all.
The time seems about right.
It looked exactly like the crash and restart on March 5th.
In other news, itās been stable overnight, despite my deleting an entire library (I moved Radio to a third Plex server, and Audiobooks will head there as well.)
well ā¦ was it 10:37 or was it not?
When was it?
What Iām saying is that I didnāt get specifically and only the dashboard unable to load error at that time.
I did get the rollerblind thing at 10:37am AZ. Which is why the log zip is at 10.43am AZ as I pulled the logs as soon as Plex made its surprise return (if Iād had to shut down Plex Script Hosts and restart PMS, it would have been a couple of minutes longer.)
OK thanks - it seems that the start of the scan of the photo library had an impact on how quickly the server responded to requests from plex Web and Plex Web must have timed out some requests for a short period
Agreed, which means it wasnāt a mini-crash as Iād thought, just the web UI losing the link and getting it back.
Thanks to everyone who have been providing logs, dumps and database copies
We have identified an issue in handling of exceptions from sqlite3 that were leading to crashes instead of errors in the log
For all users where I have seen evidence of crashes I will be sending out shortly development build 1.15.1.777 which has a fix for that and should avoid the crashes.
As it was not possible to get dumps for all of these crashes - except for those from @Wang.Chung.Plex and @wyldemusick, it is possible that some crashes may remain - but there is a high chance that the issue is the same
Please update to the new version and let me know how it goes with overnight scheduled tasks and other times. And as before, in event of any remaining issues. please gather server logs, windows Application event log export for event idās 1000 and 1001 if still crashing and any dmp files captured by windows in %LocalAppData%\CrashDumps - when you add the LocalDumps registry key to HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Windows Error Reporting
I will be sending the build to
@dirwin85 @Wang.Chung.Plex @wyldemusick @Zer0Squared @foxmen @ohmyjosh @bb1200x @mcrommert @tkg61
1 Like
Will do.
Thank you!
After installing the development build, I donāt have the crashes anymore during metadata refresh!
cool!
Iāll start a All (39) library update overnight (i use to with v1.14.x without issue, but not with v.1.15.x) and see how far it goes 
thanks - whilst we are fixing the exception crashes, db locking issues may still remain and would need to be looked into if it is causing problems - and procmon with logs is very good for that investigation - assuming db is optimized
1 Like
ah, procmon, i should then wait on the library scan until i am able to capture a filter log; as such iāll hold off scanning until tomorrow(saturday)
Let the first runs go through without any special diagnostics. If there is an issue then do specific tests as before with procmon
thank you for the quick reply! at 2am?
with your sage advice I will precede with a trial full run tonight 
