Unable to write to cache

[mark2741]

Using latest version, Unraid 6.12.13 with following plugins:

• Appdata Backup
• Community Applications
• Dynamix File Manager
• Fix Common Problems
• GPU Statistics
• Nvidia Driver
• RTL8168 Drivers
• unbalanced
• Unraid Connect
• User Scripts (with a script provided via tutorial to spin down Nvidia GPU fans automatically at array start and hourly)

 

No VMs and one docker: Plex (binhex-plex)

AMD Ryzen 2600 CPU | 16GB RAM | Nvidia 1660 Ti GPU

 

I am getting **unable to write to cache** error from Fix Common Problems. This happened last year, and I forget exactly what I did but was eventually able to 'fix' it and it was running fine until this past weekend when I couldn't access my Plex docker via GUI. I don't have a ton of setup/critical data so I just wound up trying a different docker (linuxserver-io) and it worked for maybe a day and then same issue. I then got the **unable to write to cache** and decided to just buy a new cache drive, a 250GB SATA III PCIE drive, replacing a ~5 year old equivalent. I also replaced the SATA cable with a new one. Ran fine for last ~4 days and then this morning received the error again.

 

When I try to access any of my plex libraries the UI shows "Something went wrong."

 

First, tell us the exact version of Unraid, the plugins and addons you have loaded, and what hardware you are using.  You can tell us about your hardware either here in the post, or in your signature.
 
Diagnostics attached. Thank you in advance for your assistance.

unraid-diagnostics-20240905-0813.zip

[JorgeB]

Sep  4 04:39:30 unRAID kernel: BTRFS error (device sdc1): block=1054228480 write time tree block corruption detected

 

Pool went read-only, rebooting should make it read/write again, but note that this error usually means a RAM problem, and running a Ryzen CPU with overclocked RAM like you have is known to cause data corruption, which was also being detected by btrfs, so see here and also recommend running memtest.

[mark2741]

Thank you JorgeB. I just checked and I am not overclocking the RAM per the BIOS settings, which are all just set to AUTO. I then ran memtest and immediately it started throwing errors! I then rebooted and went back into the BIOS and forced the RAM speed to 1600mhz, which is the lowest setting, restarted memtest and still got errors immediately. Fortunately, I have some old 2400mhz RAM that was sitting unused so I threw that in, reset the BIOS, rebooted and started memtest. It's running now and so far, no errors! Thank you!

 

If it caused any data corruption due to the bad RAM, do you have any next steps I should take to ensure my data is good? I'm fine with reinstalling the plex docker if needed. I have a fairly simple/small library.

Edited September 5 by mark2741

[JorgeB]

Run a scrub on the pool, but depending how the corruption happened, it my not detect everything, also any data written to the array may also be corrupt.

[mark2741]

Thanks again. Scrub ran without error. I did notice - after booting up the system for the first time after replacing the RAM - all of my apps, settings, plex docker, etc. seem just as before, but the name of server reverted back to "Tower". I changed it and all seems to be working great now.

 

I went from 16GB down to 8GB (2x4GB) but even with a plex stream running it is only using 35% of memory so I think all is well. I may buy new RAM eventually. Thanks again for your help. Fingers crossed no corruption on the array but as of now everything seems fine so far.