I can confirm it isn’t the scanner, transcoder or anything to do with the media at all. In the first two 4 AM and 5 AM, no one has been on the server to play anything since restart and it hasn’t scanned any media at all either. Take a look at the logs. I cleared them 3 times and then waited on the scheduled task to start and it dies on the same thing every time whatever it is doing right after the database backup, but the logging doesn’t say enough about what it is doing and that is with verbose and debug turned on. It is a Boost C++ error for something it is trying to load as part of one of the scheduled tasks. Once it starts getting the thread errors the main process hangs until it crashes or I have to restart it manually.
boost::thread_resource_error: An attempt was made to load a program with an incorrect format
Unless this is a false error, I normally see these in C++ when a 32-bit process is trying to load a 64-bit library or something related to that.
I have a feeling the error should be reading “Resource unavailable” or something along those lines, rather than incorrect format.
Not sure what you mean about FULL logs? That was a zip each time of everything in the log folder from application startup until it died. Is there another place I should be grabbing the logs from?
They belong to the media server executable. One other key thing that happens right before it crashes or gets those Boost errors is the memory for the process jumps up, where it normally sits around 100MB or so for normal operation, it will go up as high as 2 GB at times. I will take a screenshot next time it does it.
Oh, nm I see there are a crap ton of agent/plugin logs as well which I assume are what you are referring to.
Ok, yeah I will have to start the server back up next time and grab them that way. Not that those things tell you much and considering they are hosted in another external process it shouldn’t affect the main process anyways, but I will send them all along nonetheless. They do have Errors listed, but they are incorrectly classified as errors as the only thing I see are HTTP Errors where it couldn’t find an image or thetvdb didn’t have a particular show, etc.
This was the very first server I setup 4+ years ago. : ) It ran for a good 3 years without error before this started happening one day. I could just completely switch to Linux for Movies at least, it seems to match Windows pretty well. Linux for TV Shows is always 50 shows behind for some reason and just haven’t went to see what it was missing yet.
Ok, here are the full logs from last night. Hopefully this can shed some light on the issue with someone else taking a look though them. Thanks for the help.
OK power is back on! (well it was Friday night but we had booked a hotel already)
Trumpy here are my logs, Let me know if you see anything wonky. Thanks
@Jusedawg the next time you have a crash, do you mind zipping up the contents on C:\Users\Administrator\AppData\Local\Plex Media Server\Crash Reports\1.4.3.3433-03e4cfa35 and providing them here?
@chrisallen said:
@Jusedawg the next time you have a crash, do you mind zipping up the contents onC:\Users\Administrator\AppData\Local\Plex Media Server\Crash Reports\1.4.3.3433-03e4cfa35and providing them here?
Hi,
There are no files in those folders. I have checked there several times after a crash wanting to see if I could see what the issue was that crashed and literally nothing. It does create the folder structure though, but there are no files in them. I just assumed it was uploading them to you guys and then deleting them or something.
I went ahead and included just so you could see though for completeness.
Thanks,
Chad
@trumpy81 thank you, will look into these s well. The Detroit steel episode was badly recorded, Cable dropped out. I use no Av other than the built in win ten Av.
Added the exclusions thanks. Server has been running fine since power came back on (meaning didn’t crash for two nights and I just now added the exclusions)
Hey- Is debug loging on by default? and what is the difference between verbose, debug, and not checking either? I don’t recall ever checking debug, but it is checked.
Yesterday’s March 16th release of Plex Server version 1.4.4.3495 fixed whatever was causing my scans to stall and lock up the server. I have had to repeatedly force quit and restart the server each morning for the past week but that now has been resolved with the release of version 1.4.4.3495. I awoke to my server working just fine. The scans for both DVR EPG and media libraries completed without error and things are back to normal.
Edit: I forgot to say THANK YOU.
Surprisingly, the Plex server hasn’t crashed yet even though the ram usage is 17 times the normal usage. : )
The only time I had plex media server consume so much memory was when I had a debug setting for heap corruption investigations setting ‘enable page heap’ through glflags.exe
If it is not this, may be there is a memory leak
Might be an idea to monitor memory usage for the process through sysinternals vmmap and collect logs regularly to try and capture what was happening when memory usage climbs significantly - 2gb is the limit and when we get close failures start to occur
