Jellyfin Docker completely crashes my server

mon · Sep 4th 2023

My OMV runs rock stable. The only situation where i get problems is when using Jellyfin that i have deployed in a Docker.

This is what happens:

- Sometimes when i use to stream a Jellyfin video from a client i get a crash (not always)

- The video just plays fine for a time and then after a random time OMV crashes

- Crash means OMV is not responding anymore to anything

- There are no log entries i can find wherever i look for them about a reason

- No, it is not the RAM. I tested RAM and it is fine

- Just to have mentioned it: i mapped Jellyfin and docker to use the GPU engine of my intel CPU for rendering purpose

This situation continues since more than one year over many versions of OMV and Jellyfin.

No idea wtf is happening.

Anyone expierence the same?

Soma · Sep 4th 2023

For someone be able to help, you need to provide more info:

What hardware specs beside an Intel GPU are you using?

How are you passing the GPU to the container?

Which guide have you followed?

Which YML are you launching?

Post all needed info inside CODE boxes, please.

Don't forget to mask sensible data

raulfg3 · Sep 4th 2023

for testing pourpouse, try to do not use Intel GPU acceleration, and see if problems dissapears.

PD: Only need to comment apropiate lines in compose files (#) and "up" again

chente · Sep 4th 2023

Quote from mon

No idea wtf is happening.

It occurs to me that the directory allocated for the transcode cache might not have enough space.

mon · Sep 4th 2023

I followed the linuxserver/jellyfin docker guide and official Jellyfin documentation.

Code

root@debnas:~# lshw -short
H/W path           Device           Class       Description
===========================================================
                                    system      To Be Filled By O.E.M. (To Be Filled By O.E.M.)
/0                                  bus         N3150-ITX
/0/0                                memory      64KiB BIOS
/0/b                                memory      8GiB System Memory
/0/b/0                              memory      4GiB DIMM DDR3 1600 MHz (0,6 ns)
/0/b/1                              memory      4GiB DIMM DDR3 1600 MHz (0,6 ns)
/0/12                               memory      224KiB L1 cache
/0/13                               memory      2MiB L2 cache
/0/14                               processor   Intel(R) Celeron(R) CPU  N3150  @ 1.60GHz
/0/100                              bridge      Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series SoC Transaction Register
/0/100/2                            display     Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Integrated Graphics Controller
/0/100/13          scsi0            storage     Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series SATA Controller
/0/100/13/0        /dev/sda         disk        250GB Samsung SSD 750
/0/100/13/0/1      /dev/sda1        volume      146GiB EXT4 volume
/0/100/13/0/2      /dev/sda2        volume      9688MiB Extended partition
/0/100/13/0/2/5    /dev/sda5        volume      9688MiB Linux swap volume
/0/100/13/0/3      /dev/sda3        volume      76GiB EXT4 volume
/0/100/13/1        /dev/sdb         disk        2TB Hitachi HDS72302
/0/100/13/1/1      /dev/sdb1        volume      1863GiB EXT4 volume
/0/100/14                           bus         Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series USB xHCI Controller
/0/100/14/0        usb1             bus         xHCI Host Controller
/0/100/14/0/1                       bus         ASM107x
/0/100/14/0/5                       bus         USB2.0 Hub
/0/100/14/1        usb2             bus         xHCI Host Controller
/0/100/14/1/1                       bus         ASM107x
/0/100/1a                           generic     Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series Trusted Execution Engine
/0/100/1b                           multimedia  Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series High Definition Audio Controller
/0/100/1c                           bridge      Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series PCI Express Port #1
/0/100/1c.1                         bridge      Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series PCI Express Port #2
/0/100/1c.2                         bridge      Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series PCI Express Port #3
/0/100/1c.2/0      eth0             network     RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller
/0/100/1c.3                         bridge      Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series PCI Express Port #4
/0/100/1c.3/0      scsi2            storage     ASM1062 Serial ATA Controller
/0/100/1c.3/0/0    /dev/sdc         disk        2TB WDC WD20EFRX-68E
/0/100/1c.3/0/0/1  /dev/sdc1        volume      1863GiB EXT4 volume
/0/100/1c.3/0/1    /dev/sdd         disk        2TB WDC WD20EFRX-68E
/0/100/1c.3/0/1/1  /dev/sdd1        volume      1863GiB EXT4 volume
/0/100/1f                           bridge      Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx Series PCU
/0/100/1f.3                         bus         Atom/Celeron/Pentium Processor x5-E8000/J3xxx/N3xxx SMBus Controller
/0/1                                system      PnP device PNP0b00
/0/2                                system      PnP device PNP0c02
/0/3                                system      PnP device PNP0c02
/0/4                                system      PnP device PNP0c02
/0/5                                system      PnP device PNP0c02
/0/6                                system      PnP device PNP0c02
/1                 vethc64e5e85     network     Ethernet interface
/2                 br-66411dfe4399  network     Ethernet interface
/3                 br-b57450777507  network     Ethernet interface
/4                 br-c97b03a534b8  network     Ethernet interface
/5                 docker0          network     Ethernet interface
/6                 vethfa2e40a      network     Ethernet interface
/7                 veth79446dd      network     Ethernet interface
/8                 vethce24027      network     Ethernet interface
/9                 vethb720da6      network     Ethernet interface
/a                 veth1ef22b6      network     Ethernet interface
/b                 vethed003f5      network     Ethernet interface
/c                 vethfde3a16      network     Ethernet interface
/d                 vethd601c25      network     Ethernet interface

Display More

I am not using a YAML but did the config myself with Portainer.

Here are the probably relevant parts including the GPU passthrough.

Let me know if you need anything else

mon · Sep 4th 2023

Quote from chente

It occurs to me that the directory allocated for the transcode cache might not have enough space.

This also happened when not transcoding

But because you asked for it: currently 17,5 GB are free on this drive

mon · Sep 4th 2023

Maybe the Jellyfin part is also relevant:

chente · Sep 4th 2023

Quote from mon

This also happened when not transcoding

So I'm sure that's not the reason.

Quote from mon

Here are the probably relevant parts including the GPU passthrough.

You said that you use the integrated Intel GPU, however in that image you posted something related to NVIDIA appears. what's that?

chente · Sep 4th 2023

Quote from mon

Maybe the Jellyfin part is also relevant:

Does your GPU support all those codecs?

You can check it here. https://es.wikipedia.org/wiki/Intel_Quick_Sync_Video

mon · Sep 4th 2023

Quote from chente

Does your GPU support all those codecs?
You can check it here. https://es.wikipedia.org/wiki/Intel_Quick_Sync_Video

If it wouldn't then the video would not start, right?

mon · Sep 4th 2023

Quote from chente

So I'm sure that's not the reason.

You said that you use the integrated Intel GPU, however in that image you posted something related to NVIDIA appears. what's that?

Sorry i don't see anything Nvidia. What are you referring to?

chente · Sep 4th 2023

Quote from mon

If it wouldn't then the video would not start, right?

No idea. In any case, I don't think it's good to tell Jellyfin to hardware transcode a certain format but then the system is not capable of doing so. I don't know what could be the result of doing that.

Quote from mon

Sorry i don't see anything Nvidia. What are you referring to?

The last line of the left column.

mon · Sep 4th 2023

Ok fair enough. I changed the encoding options accordingly:

Regarding NVIDIA_DRIVER_CAPABILITIES

I think this is not something i have added but must be standard of the docker container.

Found somewhere on Github about it: "This option controls which driver libraries/binaries will be mounted inside the container."

Might not be relevant. But i can try to get rid of this and redeploy the container.

chente · Sep 4th 2023

Quote from mon

and redeploy the container

I would do it through a compose file, I think it is faster and easier to manage. You can still do it in Portainer if you want, although I recommend using the plugin.

Soma · Sep 4th 2023

Output of:

docker inspect jellyfin

And have you tried to use Intel Quick Sync instead acceleration of VA-API?

According to this, your processor supports it.

As for beeing created by hand with portainer, you can make a YML from it, just to keep it save by running: change <container_name or ID> with the proper one

Code

docker pull red5d/docker-autocompose && docker run --rm -v /var/run/docker.sock:/var/run/docker.sock red5d/docker-autocompose <container_name or ID>

Other than for you to make sure that all the settings are supported by your Hardware, I can't tell you much more.

I will only have to deal with this situation once my Storaxa arrives which is the one with Intel CPU/GPU.

For now, I only run Jellyfin with V4L2 acceleration (CM4)

mon · Sep 4th 2023

Quote from chente

I would do it through a compose file, I think it is faster and easier to manage. You can still do it in Portainer if you want, although I recommend using the plugin.

Yeah well i use it sometimes. But honestly i doesnt change much. I am used to doing this in Portainer. It is more or less exactly as easy as using YAML. In this case i can just change the respective value and redeploy.

The advantage of YAML to me is that you can copy paste and alter existing YAML files you get from internet.

Also if you have multiple containers in 1 stack.

mon · Sep 4th 2023

Quote from Soma

Output of:
docker inspect jellyfin

mon · Sep 4th 2023

Quote from Soma

And have you tried to use Intel Quick Sync instead acceleration of VA-API?
According to this, your processor supports it.

Yeah i have been testing that at some points way back but decided to go with VA-API due to a reason i cannot remember.

Although after this post of chente the VA-API driver is just a driver for Quicksync??

Thread

How to activate Intel Quick Sync in docker (Jellyfin, Handbrake,...)

Intel Quick Sync is a hardware video accelerator built into some Intel CPUs. It allows converting the video format in real time without overloading the CPU. It is useful for media players or video converters.

Prerequisite:

Make sure the CPU has Intel Quick Sync technology. Check the following information:

https://en.wikipedia.org/wiki/Intel_Quick_Sync_Video
https://ark.intel.com/content/…873&0_QuickSyncVideo=True

Install the drivers:

Choose the right driver for our CPU. See here which one you need: …

chente

Apr 5th 2021

chente · Sep 4th 2023

Quote from mon

Yeah well i use it sometimes. But honestly i doesnt change much. I am used to doing this in Portainer. It is more or less exactly as easy as using YAML. In this case i can just change the respective value and redeploy.
The advantage of YAML to me is that you can copy paste and alter existing YAML files you get from internet.
Also if you have multiple containers in 1 stack.

Just for the ability to make a simple copy of that compose file into a text file and save it somewhere I think it's worth the switch. On the other hand, the plugin makes that copy for you automatically.

mon · Sep 4th 2023

Quote from chente

Just for the ability to make a simple copy of that compose file into a text file and save it somewhere I think it's worth the switch. On the other hand, the plugin makes that copy for you automatically.

Ok great, then i will do nothing

Participate now!