NVME cache pool errors while transcoding plex

[qnzkrew]

updated to  6.11rc4 with i-5 12600k alder lake cpu and finally got transcoding working but as soon as i try to transcode multiple streams i get cache btrfs errors and shuts down everything running off the cache pool. cache pool is twin crucial 1tb nvme in raid 1 .

 

this is the relevant part of diagnostic and i also uploaded full diagnostic below 

https://pastebin.pl/view/2df7ed41

tower-diagnostics-20220909-1804.zip

[trurl]

Have you done memtest?

[JorgeB]

Already recommended that to the OP, did you do it?

[qnzkrew]

5 hours ago, JorgeB said:

Already recommended that to the OP, did you do it?

im sorry i missed your post. if its not too much of a hassle. how would i run memtest on unraid. thank you and sorry again for not oroginally seeing your reply

 

EDIT: 

im guessing i can just pick memtest86 from the bios startup. one question , does mem test only test ram memory. if so, do you guys think that the btrfs errors in the nvme is caused by ram or does the memtest also check nvme . thanks again 

Edited September 10, 2022 by qnzkrew

[icedfusion]

I would highly recommend that you change your transcoding from being on your nvme disk to being in RAM - this can be easily done by changing your plex docker config for the path /transcode to /dev/shm.

Really, if you share your plex server and it is used, the number of read/writes is large and you really should be doing this from RAM and not your NVME disk.

It could solve your issue too.

 

[edit]

to run memtest, when unraid is booting you can stop the boot into unraid and select a memtest. Also, there is generally a memtest faciliy as part of your bios

 

Cheers

ice.

Edited September 10, 2022 by icedfusion
answer memtest question

[qnzkrew]

10 minutes ago, icedfusion said:

I would highly recommend that you change your transcoding from being on your nvme disk to being in RAM - this can be easily done by changing your plex docker config for the path /transcode to /dev/shm.

Really, if you share your plex server and it is used, the number of read/writes is large and you really should be doing this from RAM and not your NVME disk.

It could solve your issue too.

 

[edit]

to run memtest, when unraid is booting you can stop the boot into unraid and select a memtest. Also, there is generally a memtest faciliy as part of your bios

 

Cheers

ice.

okay i will check into this as soon as i finish running the tests . thank you 

[JorgeB]

14 hours ago, qnzkrew said:

im guessing i can just pick memtest86 from the bios startup

Yep, as long as you boot CMS/legacy, it won't work with UEFI boot.

 

14 hours ago, qnzkrew said:

do you guys think that the btrfs errors in the nvme is caused by ram or does the memtest also check nvme .

It only checks the RAM, but those errors appear to be the result of bad RAM