unraid 7.3.0 RAID5 total capacity reporting incorrect causing storage used percentage to report incorrectly

[soulvoid86]

Describe the bug
Total capacity showing incorrectly for RAID5 arrays in Unraid 7.3.0

To Reproduce
Configure a secondary array with RAID5 (allocation profile set to raid5 on creation).
Use this as the cache drive in PlexCache-D

Expected behavior
This may be a due to a bug with unraid. PlexCache shows both the free and available space correctly, but the Total Capacity shows all 3 drives. Thus, the percentage used reports incorrectly as it seems to be reading from capacity.

Screenshots

Desktop (please complete the following information):

• OS: Win11 Pro
• Browser: Chrome
• Version: 149.0.7827.114

Additional context
I know unraid should correct this on their side, but in the meantime, is there a way to calculate the percentage free differently maybe?

[Brandon-Haney]

Thanks for the detailed report. I'm fairly sure this is a btrfs RAID5 reporting quirk rather than an Unraid 7.3 regression. On btrfs RAID5/6 pools the kernel reports total capacity as the raw sum of all member drives (your 3 disks) instead of the usable post-parity size, but free/available come back correct. PlexCache reads total from the standard statvfs call, so it inherits that inflated number and the used % ends up wrong, which lines up with what you're seeing.

There's already a setting that fixes this without needing an update. It was originally added for ZFS pools but it works the same way here:

Settings > Cache > Storage Limits > "Cache Drive Size". Set it to your real usable RAID5 capacity (e.g. 7.3TB). PlexCache keeps the accurate free-space reading and recomputes used as total minus free, so the percentage corrects itself. Leave it blank to go back to auto-detect.

I'd also like to see if I can make this auto-detect properly so the override isn't needed. The catch is the container can't always reach the same disk tooling the host has, so I need to confirm what PlexCache can actually see from inside it. If you don't mind, could you run these and paste the output?

On the Unraid host (Terminal), swap /mnt/cache for your RAID5 pool's mount if it's different:

df -h /mnt/cache
btrfs filesystem usage /mnt/cache

Inside the PlexCache container (swap plexcache-d for your container name):

docker exec plexcache-d sh -c '
  echo "--- df ---"; df -h /mnt/cache;
  echo "--- python statvfs ---"; python3 -c "import shutil; print(shutil.disk_usage(\"/mnt/cache\"))";
  echo "--- btrfs binary? ---"; command -v btrfs || echo "no btrfs binary";
  echo "--- btrfs usage ---"; btrfs filesystem usage /mnt/cache 2>&1 | head -20;
  echo "--- sysfs ---"; ls /sys/fs/btrfs/ 2>&1
'

The container output is the one that matters. If any of those shows the true usable capacity from inside the container, I can wire up auto-detection so you won't need the manual override. If not, the Cache Drive Size setting above is the way to go.

[soulvoid86]

Perfect, that's good enough for me. I know this is an issue at the host level with Unraid, not just 7.3. I just found PlexCache and am setting it up for the first time.

Hopefully Unraid fixes this themselves so devs like yourself don't need to bother. I know it's an experimental feature as well, so hopefully they get there.