Hello all!

 

I'm at my wit's end here, trying to diagnose for weeks. I'll try my best to describe the situation. I have had unraid going strong for a couple of years now, but these last two months something has changed. I don't remember the initial cause, since I didn't notice the disk activity, but for some reason, my cache was completely full, and the disks were writing to each other. Moving the data off the cache manually helps, but then within 24 hours, it fills up again, once full the disks write to the emptiest disk, and once full continues to the next. Based on logs I'm pretty sure it's not mover's fault, I've tried:

Turning off V.M's

Turning off Docker, which I couldn't tell if it worked

Several restarts

Updated to the latest Unraid

Sonarr was throwing corrupt errors, fixed with an old backup

 

The weirdest part is that all my shares have remained unchanged, with the correct transfer of data, i will link it in images. But the Tower still moves media ONTO the cache. I hope this is enough info, and I'll be around today if you need more information as well. 

 

Thanks for the help.

tower-diagnostics-20250428-0846.zip tower-syslog-20250428-1445.zip

Can you post a screenshot of the share settings for the problematic share?

 

because accord to this the only shares that should remain on dram_cache pool are the unraid default shares, appdata, domains, isos, and system. Everything set to shareUseCache="yes" should be moved to the array by mover.

 

It does look like the shares on the bottom of the list should not exist on the cache pool and data does, so that data should be moved off.

 

appdata                           shareUseCache="prefer"  # Share exists on dram_cache
d--a                              shareUseCache="yes"     # Share exists on dram_cache, disk1, disk2, disk3, disk4
domains                           shareUseCache="only"    # Share exists on dram_cache
e-------n                         shareUseCache="no"      # Share exists on disk1, disk2, disk3, disk4
g---s                             shareUseCache="yes"     # Share exists on disk3, disk4
isos                              shareUseCache="prefer"  # Share exists on dram_cache
system                            shareUseCache="prefer"  # Share exists on dram_cache
Y-T                               shareUseCache="prefer"  # Share does not exist
W--------0                        shareUseCache="no"      # Share exists on no_dram_cache
W-------0                         shareUseCache="no"      # Share exists on no_dram_cache

 

Edited April 28, 20251 yr by MowMdown

Sure! here is the attached screenshot for the data share, which is the only share moving constantly. The two shares at the bottom i just deleted, they were empty windows shares that i was going to use but never did. Hopefully this helps, again thank you for helping i appreciate it!

Click on your disk number on the main dashboard page, what is the minimum free space on each disk setting? how full is your array?

 

Disregard, array disks don't seem to have a minimum free space setting. Only pools.

 

I see a lot of "share cache full" and "error: No space left on device" messages which to me looks like you are hitting a minimum free space setting on the disks themselves but not the share.

Edited April 28, 20251 yr by MowMdown

do you have any recomendations? its just so weird that out of nowhere it started doing this, no changes to shares or disks!

 

**edit**

I adjusted the min free space on each pool at least, if that helps?

Edited April 28, 20251 yr by AMPOPTART

I just don't see anything out of the ordinary based on the information you've provided. New files written to the data share should be written to the dram_cache first and then should be moved to the array based on the screenshot you've shared.

You do have mover scheduled to run nightly correct?

Logs are useless because they're just full of "share full" and don't include boot. Reboot and post diags again.

Do you have a Mover Tuning plugin installed? If so that can be set to fill the cache to a certain level which if set wrong could be going against what you want.

Edited April 28, 20251 yr by Kilrah

5 hours ago, AMPOPTART said:

The two shares at the bottom i just deleted, they were empty windows shares that i was going to use but never did.

According to those earlier diagnostics, those 2 shares were the only shares actually using the pool named no_dram_cache. But you say they were empty. Why then is no_dram_cache nearly full?

 

I suspect you have files at the top level of one or both of these pools, and so not actually in any user share, and so not going to be moved.

 

What do you get from command line with this?

ls -lah /mnt/dram_cache

and this?

ls -lah /mnt/no_dram_cache