Whoops, was looking in the wrong section, not under “Windows Logs”.
Sending this via Private Message as well, but I am seeing a ton of errors with another program, which directly connects to Plex, so that feels like a possible area of contention.
Sorry, forgive my ignorance, but which files are 0 bytes? None of the files listed in the zip i sent have 0kb, and i’m not sure which other files you’re after. here is my event viewer logs for the past 7 days. upload for plex.zip (171.9 KB)
I quoted the wrong post !!! It was @quik3115
sorry I posted this to wrong post earlier - putting it right now:
Thanks for sending the zipped dmp files. As I suspected the dmp files are empty - file size = 0 bytes
It is not going to be easy to fix these crashes blindly without dumps
Lets see if there are clues in the windows application event log - on type of crashes. And always update to latest version of Plex Media Server and we can see then if things change and if you get more logs of the crashes, I can see if there is a pattern of what activity before crashing
Apologies for responding to your post in error earlier.
I have looked at the logs and also the uploaded crash reports.
All the crashes relate to hardware transcoding. I am seeing two types of crashes. Both are mentioned in this post which consolidates the results of the hardware acceleration related crashes investigations that I did.
I can see crashes in the intel hardware graphics driver dll igdumdim32.dll - these are similar crashes to those in bullet 1 in the linked post. The crashes were resolved by switching to different version of the intel driver. See the table in bullet 1 which was put together from users feedback
I am also seeing a crash that is similar to what is covered by bullet 3 in my post. That was seen when Plex Media Server was run headless or through RDP or as a service ot windows scheduled task.
Awesome! Inshallah, updating that driver resolves it! Thank you for your help!
I can look through the other crash dumps that I saw in that directory and see if there’s anything else of value in there.
I’ve updated to the latest version of Plex Media Server that looks like it was released today. We’ll see if anything changes there.
Should I enable Verbose logging for the time being?
if we are only after capturing logs for these crashes - then yes verbose logging may add more clues, But if you need to get other issues looked into then verbose logging would not appropriate
Thanks for the event log exports. Could only find two Plex Media Server crashes - from 17th and 16th March.
Crashes were
0xC0000409 STATUS_STACK_BUFFER_OVERRUN
The system detected an overrun of a stack-based buffer in this application. This overrun could potentially allow a malicious user to gain control of this application.
I think we just need to see more logs to see if there is some consistency of what was happening before a crash
No crashes since the latest PMS update… I’ll update with new logs again if I get another crash.
Just got another crash, restarted PMS, immediately gathered log data, uploading below.
Plex Media Server Logs_2020-03-27_20-25-33.zip (4.5 MB)
I also checked the Crash Reports folder, and do not see any new crash reports.
I’m at the point now that I want to try something that may fix this issue. Should I reinstall fresh? Should I wipe out some library metadata stored somewhere? I mean… I’m all about finding the root cause and helping the Plex team debug, but I feel like I’m not able to provide you any valuable debug data.
Any suggestions?
And another crash…
We have some clues now - we are running out of memory - there may be a memory leak. The memory allocation errors started after running for about 30 hours in the last crash - with the first at this time
Apr 05, 2020 19:24:06.558 [2744] ERROR - Thread: Uncaught exception running async task which was spawned by thread 16508: bad allocation
There were more of these exceptions until eventually crashing and I suspect due to same issue - shortage of memory
Memory leaks are not easy to troubleshoot - but we can start by watching memory use. So I suggest you download SysInternals VMMap and launch it and connect it to the Plex Media Server.exe process very soon after launching Plex Media Server. So launch Plex Media Server afrresh and immediately connect to the process from VMMap. Then get VMMap to display timeline and set the option to have refresh every 10 seconds.
Keep it running and we see memory usage exceed 1Gb - suggest capturing Plex Media Server logs then and also take a screenshot of VMMap screen save VMMAP data into mmp file an
Continue to have VMMap connected to the Plex Media Server.exe process and monitoring memory use. And when usage exceeds 2Gb - repeat what you did at 1Gb.
When viewing VMMap data, it is best to force a refresh before taking any screenshot
Repeat again as it gets to 3Gb - PMS would probably crash when we get close to 4Gb
As it gets to 3.5 Gb - if you spot that or anytime after 3Gb - we could get screenshot of vmmap, save vmmap data and force a Plex Media Server process dump before it crashes itself
Zip these sets into separate zips - each with plex media server logs zip and upload to dropbox / google drive etc and send me link by Private Message
I’ve been watching the VMMap memory visualization as much as I can while the server is online. I don’t use this PC very often, so I’m not monitoring it throughout the day.
I went about 5 days without a crash, and I thought all had been magically fixed. However, in the last two days, I have had 2 crashes again.
I am uploading these logs. If you would like me to try another VMMap profiling, please let me know, but I will say it is very hard to catch the problem through VMMap. I am yet to see any memory go over the ~215MB threshold.
Plex Media Server Logs_2020-04-27_20-42-14.zip (3.7 MB)
Plex Media Server Logs_2020-04-29_08-59-39.zip (4.2 MB)
Again, any help would be appreciated, or if you think it’s safe to just reinstall Plex at this point, I’d rather do that then constantly have my server crash.
For the last crash the memory allocation errors started at 03:24 am on 29th April after running for 27 hours.
For the earlier crash it was after running for about 50 hours
when you notice that it crashed overnight - I thought vmmap would still have the history if it is left with the auto refresh option. Have you saved any vmmap data into mmp ? Of course if you refresh vmmap after the crash it would show zero memory but may still have the timeline and data for the time before that
I will see if we can add some extra logging
Thought I got past the random crashes by updating a bunch of firmware, GPU drivers, overall cleanup, etc… Had a run for a few days, and then again… a new crash.
Plex Media Server Logs_2020-05-09_15-15-30.zip (4.2 MB)
The unique bit around this crash, I was actively attempting to watch Plex on my FireTV (Gaming Edition). I attempted to rewind a bunch of times at the beginning of a playback (likely beyond the initial start time - ie I was 40s in, but rewinded 150s back), after this Plex suddenly stopped responding and crashed.
I have no idea if this is related, and I can attempt to reproduce, or it was just a wacky coincidence.
I will restart the Plex Server again, and try to use VMMap against it, but again, when it’s running for multiple days, I think there’s problems with using the VMMap tool.
Same as before.
Memory allocation failures started at this time
May 09, 2020 11:48:35.821 [7848] ERROR - Thread: Uncaught exception running async task which was spawned by thread 16936: bad allocation
and eventually crashed after this allocation failure
May 09, 2020 11:50:08.849 [16932] DEBUG - CERT: incomplete TLS handshake: malloc failure
PMS was launched at May 07, 2020 19:44
I will try and see if I can get you a build that would log memory usage in the server log - that would be a better option than VMMAP.
Received another crash. VMMap Timeline never exceeds ~205MB used, so it doesn’t look like an overflow issue. Sending the .mmp in a private message.
Plex Media Server Logs_2020-05-14_20-39-22.zip (4.6 MB)
[Plex Media Server Logs_2020-05-14_20-39-22.zip|attachment]
I think there’s an issue with VMMap that it doesn’t record timeline values unless you manually press the View-> Refresh button. So, if there is an overflow error, it may not have been captured? Maybe the .mmp file has that info? Looking at it, it’s only about 3MB… doesn’t seem like it holds anything valuable.
Either I’m not using this correctly, or when the Plex process crashes, and you do a View Refresh in VMMap, it loses any trace data?
Thank you - I am sending you a development build which logs memory use on every request so no need for VMMAP - hopefully we will get closer to the problem area
Does anyone have a link where I can download 1.18.6.2368 Plex Server for Windows?
The latest version keeps crashing and I’ve gotten no response on the forum from anyone at Plex.
Thanks in advance,
Jim
