Random crashes on DS216+II, DSM 6.2.3, PMS 1.22.3.4392 and earlier

Synology DS216+II
DSM 6.2.3-25426 Update 3
I had crashes on 1.21.1.3876, updated to 1.22.3.4392, and had another crash, so I decided to post the logs.

Crashes happen once in a few days, sometimes at night, sometimes during the day. I created Tautulli “Plex Down” notification to know when it happens.

I do not have on/off schedule on my DSM.
Plex Scheduled Tasks are set to 3pm and 6pm.

Tautulli notified me about Plex stop at 3:20 AM (on 1.22.3). No one used Plex at the time, so it must be a result from some background job. Also there’s nothing in the history of Sonarr/Radar during that time, so no new content was added.

Before the crash I had “Enable Plex Media Server debug logging” checked, and verbose set to off.

Manual archive of the folder before the Plex start: Logs 240421.zip (9.6 MB)

Downloaded after the Plex start from UI: Plex Media Server Logs_2021-04-24_15-15-43.zip (5.4 MB)

Just in case, now I disabled “Send crash reports to Plex” so I’d have a DMP file instead in the future. For now I’m waiting for another crash to post it.

Please let me know when you have a dmp file.

Put it in a ZIP then attach it here.

I will see if I can track it down to something.

1 Like

I’d appreciate the assistance, thanks!

It happened again today (25.04) around 18:58. Few hours before the crash I noticed DSM was using CPU (indicator in the red, with the Plex task on top of sorted tasks). ~4 hours after the crash I went to start Plex back, but noticed the CPU was still being used, went to see what’s using it - and it was Plex, even though its package was stopped as of few hours. It’s not supposed to be that way, right?

PMS DMP file:
1.22.3.4392-d7c624def_PMS_Crash_Repor_DMP.zip (74.3 KB)

Logs: Plex Media Server Logs_2021-04-25_22-25-59.zip (4.9 MB)

Thank you.

I am looking at a valid dump file with traceback information.

I don’t recognize this part of the server code so don’t know what it is.

I am writing it up for engineering to look at it.

1 Like

Thanks!

After that crash I started the PMS, and after my message here I opened iOS client (latest beta), and PMS crashed again (at 22:36) while the client was loading the content. I restarted Synology after that, Plex launched itself and I was able to watch from iOS client no issue. Added DMP for the new crash:

1.22.3.4392-d7c624def_PMS_Crash_Repor_DMP_250421_2236.zip (65.7 KB)

Plex Media Server Logs_2021-04-25_23-15-37.zip (4.2 MB)

Crashed again today (26.04.21) at 16:18. Wasn’t using it at the time.

1.22.3.4392-d7c624def_PMS_Crash_Repor_DMP_260421_1618.zip (69.5 KB)

Plex Media Server Logs_2021-04-26_16-53-43.zip (3.2 MB)

@enchained

I’m starting to become very suspicious of your Syno / installation there.

What you’re experiencing is not the norm.

Are you willing to conduct an experiment with me?

I can try, please tell me what to do.

I think it crashed much less frequently on 1.21.1.3876 (and wasn’t crashing at all on some earlier version), now after the update, it became daily. But certainly, DSM is a part of it too. I know I’m not on its latest version, but I can’t update yet because some breaking API changes were introduced and I have to wait when it’s fixed in every community software that uses it.

This test sets up a new, temporary, “Test” server while keeping your existing server safe. It will help us figure out if the problem is somewhere in the PMS database or in DSM itself.

Here’s what I’d like you to do

  1. Make certain latest PMS is installed

  2. Stop PMS

  3. Sign out of the Plex/web app (upper right corner), close the Plex/Web tab.

  4. Open the Plex shared folder in FileStation.

  5. Rename Library to Library.save (we’re going to keep what exists safe)

  6. Start PMS

  7. Let it startup for a bit (about a minute)

  8. Open an Incognito window to http://ip.addr.of.syno:32400/web

  9. Start to sign into this new server “Got It”.

  10. THIS IS IMPORTANT

  11. Give this new server a Friendly name which is unique - “TEST SERVER” (we don’t want the name to collide with the default name for the Syno)

  12. After you have “CLAIMed” this new “TEST SERVER”, you no longer need to use the Incognito window.

Here’s where the test will take some time to complete.

  1. Define a library section and let it populate before doing anything.
  2. Use it and see if you can recreate the crash. Be certain to allow enough time
  3. If you cannot make it crash, add another library section
  4. Repeat steps 2 & 3 until everything is fully built back up.

Possible outcomes are:

  1. PMS crashes even in this test configuration → DSM / host problem

  2. PMS does not crash after a complete rebuild of the entire configuration → PMS problem.

How do you feel about this?

Thanks, I think I will be able to try it around the weekend. Does it matter if the libraries I create use the exact same agent etc? Is it better to write down my server and each library settings to recreate exactly, or it’s better to create a mostly default config?

Also, is there any use in uploading new dumps and logs from the current server, until I will be able to conduct the experiment?

It’s perfectly OK (actually desired) to recreate this test server as you would normally create it. It’s important to create a controlled test case while keeping track of everything done as it’s done. We need to figure out how we arrive at the failure point

1 Like

I had an idea to check /var/log for times when dumps were created. I found out that each crash has a corresponding line in the kern.log:

2021-04-25T18:57:02+03:00 DiskStation kernel: [688897.142755] Plex Media Serv[12423]: segfault at 0 ip 000000000191b40c sp 00007f39025a02d0 error 4 in Plex Media Server[a30000+f5b000]

2021-04-25T22:36:01+03:00 DiskStation kernel: [702036.101081] Plex Media Serv[11689]: segfault at 0 ip 000000000191b40c sp 00007f75276c02d0 error 4 in Plex Media Server[a30000+f5b000]

2021-04-26T16:18:30+03:00 DiskStation kernel: [63514.354013] Plex Media Serv[27817]: segfault at 0 ip 000000000191b40c sp 00007f37177342d0 error 4 in Plex Media Server[a30000+f5b000]

2021-04-28T15:35:03+03:00 DiskStation kernel: [233705.269365] Plex Media Serv[31643]: segfault at 0 ip 000000000191b40c sp 00007f3187a222d0 error 4 in Plex Media Server[a30000+f5b000]

2021-04-28T23:20:09+03:00 DiskStation kernel: [261610.109185] Plex Media Serv[21697]: segfault at 0 ip 000000000191b40c sp 00007f1b774462d0 error 4 in Plex Media Server[a30000+f5b000]

2021-04-29T05:00:26+03:00 DiskStation kernel: [282027.021927] Plex Media Serv[8842]: segfault at 0 ip 000000000191b40c sp 00007f8fa3d102d0 error 4 in Plex Media Server[a30000+f5b000]

2021-04-29T16:21:37+03:00 DiskStation kernel: [322897.805329] Plex Media Serv[26050]: segfault at 0 ip 000000000191b40c sp 00007f72794c32d0 error 4 in Plex Media Server[a30000+f5b000]

https://utcc.utoronto.ca/~cks/space/blog/linux/KernelSegfaultMessageMeaning

With a faulting address of 0 and an error code of 4, we know this particular segfault is a read of a null pointer.

Do you have more info on that? Could you relay that to the engineering team if possible?

I’ve never experienced segfault in Linux before, so I’m not sure what to do with it. Is the issue specific to some address in the Plex code, or should I also contact Synology support? Maybe I should do a memtest?

@enchained

Is this on the new test server?

Didn’t have the time to set it up yet, so it’s all past crashes on the regular server, including latest one. Does segfault specifically imply if the problem is with Plex or the host? Is the experiment still required to figure this thing out?

It’s usually a software issue but I have seen cases where it’s the OS instead of the application.

In a couple, more rare, instances, it’s actually been the hardware.

That’s why I would like to have a known-good PMS basis to work from.

If PMS is known fresh, and it crashes, then we look to DSM / hardware and Synology

1 Like

Since I have 10+ libraries and it could be up to 1+ day without crashes, by my calculations the experiment would take me at least a couple of weeks. So before starting it, I decided to try IFYates advice from here just in case. And it probably worked, at least I had no crashes for 5 full days (before it was almost every day).

The only place I could find info about the proper reinstall steps is here:

Since it’s titled as “FAQ 8: Uninstalling PMS - Reverting to a previous version” and I didn’t need to downgrade, I didn’t read this before. And I’d guess a lot of users also don’t try to reinstall it that way cause they’re too afraid of the “Uninstall” action, thinking it will remove their data or settings.

I think it would be useful to create a separate FAQ entry for “How to update/reinstall”, or add it as a troubleshooting step. But, in case similar issues may happen with the future versions, I think the best place to show a clean update routine on Synology would be here: https://www.plex.tv/media-server-downloads/#plex-media-server

Going forward,

You can rest assured, I take every precaution to safeguard users’ data on Synology

When you eventually migrate to DSM 7, you’ll see how much I do in the new packaging.

I keep everything in different areas just as it is in DSM 6.
The new menu mechanism gives the opportunity to confirm as well as ‘back up & change’ before making the final ‘commit’ action.

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.