Hello all,

been having some strange issue with one specific HDD in my cache only writing at ~140 MB/s and thus dragging down writes to the entire cache.

I've run a test with iperf to ensure I haven't somehow dropped to gigabit ethernet speeds.

Next I ran hdparm to check writes to the individual drives in the cache and confirmed that one drive in particular is being slow

Also confirmed the issue using a plain ol' file transfer over the network.

Other things I did: Rebooted and updated Unraid, move sdh drive from its own Sata connector to my HBA, checked manufacturer speck to make sure it can actually do the usual SSD speeds, checked SMART and checked everything is plugged in correctly

None of which showed any significant change in write speeds.
Interestingly, reads from the cache are fine, giving me some 700MB/s in windows

The share is set to use Cache mainly and only fall back to the array should it fill up. Current utilization is 52% and minimum free space is 10 MB so this shouldn't cause this issue either.

Sleuthing the forums and search engines hasn't really given me any answers either which has brought me here to ask if anyone else here has had issues like this.

As is customary I've also attached a syslog in case it helps.

tower-syslog-20250709-1635.zip

Please post the diagnostics.

2 minutes ago, JorgeB said:

Please post the diagnostics.

Whoops, forgot to include those. Here they are tower-diagnostics-20250709-1853.zip

Nothing obvious logged, likely just a slower SSD for writes, not all SSDs can sustain high write speeds for more than a few GBs, though I'm not familiar with that Sandisk model, could also just be getting tired.

4 minutes ago, JorgeB said:

Nothing obvious logged, likely just a slower SSD for writes, not all SSDs can sustain high write speeds for more than a few GBs, though I'm not familiar with that Sandisk model, could also just be getting tired.

Hm, the file was only 2 GB in size but I guess it being a pretty budget tier SSD it might just be that. Would certainly explain the very brief burst of higher writes at the start.

As for it getting tired, the drive isn't that old yet, bought it about a year ago. Not entirely sure how getting hammered by unraid would affect the lifespan though.
By comparison, the crucial drive is about half a year older and still doing fine at least in terms of writes.

All in all, I guess a solution would be to get a new drive and test with that after removing the old drive from the array. Wouldn't be too inconvenient as I do need a new drive for my main PC anyway.