Unraid 7.1.4. I do monthly scheduled parity checks. Check on 8/1 ran without error. A few days later I ran a user script to clean up but docker image file, which also ran without error. I decided to run the parity check again because I've been wondering if the docker cleanup script causes errors in the next scheduled parity check. (I don't run the script often enough to know. Just a theory I wanted to check.) During the follow-up manual parity check, one of the disks reports a number of read errors and it was removed from the array. Not the drive with the docker image file.

I ordered a pair of new WD Red drives to match the size of my parity drive. Figured I might as well upsize the broken drive. I precleared one of the WD Red drives via an external USB cradle. I did one preclear pass and had no problems. I replaced the failed drive with the new WD Red, and the rebuild started. About 30 minutes into the rebuild, the replacement drive reported errors and left the array.

I decided to run another preclear pass on the replacement drive while it was inside the case and connected via SATA. That preclear completed without error.

This morning I added the replacement drive back into the array and the rebuild started. Again there were errors about an hour after I started the rebuild. I opened the server up and checked cables. This drive is connected via a HBA I added. I made sure the cable to the HBA was well connected. I re-seated the card. I decided to move the replacement drive to another port on the same HBA. As I write this, I'm running a 2-pass "erase and clear" preclear.

I'm puzzled as to why the preclears all seem to run without a problem.

If someone has thoughts or other things to consider, I'm all ears.

hulk-diagnostics-20250811-0547.zip

Edited August 11, 2025Aug 11 by Abe677
Added diagnostics file.

Please post the diagnostics.

Sorry. Just added the file to my original post. I don't spend much time in this forum. Unraid has been trouble free for about 5 years for me.

SMART for disk4 looks fine, and it just dropped offline before; this is typically a power/connection issue.

The emulated disk is mounting. Assuming contents look correct, recommend checking/replacing cables and rebuilding on top:

https://docs.unraid.net/unraid-os/manual/storage-management#rebuilding-a-drive-onto-itself

Thanks for the advice so far. I had moved the drive to another port on the HBA, then started a "erase and clear". I had originally set this for 2 passes, but stopped during the post-read on pass 1. No problems during the 45+ hours of pass 1.

Last evening I brought the array up in maintenance mode and started the array sync. About 2 hours in the drive logged more read errors and the sync stopped. I stopped the array. I'm attaching another diagnostic file.

Next I'm going to shut the server down , open it up again, and check cabling. I may move the drive power to another source, then try again.

hulk-diagnostics-20250813-0747.zip

Edited August 14, 2025Aug 14 by Abe677
typo

Disk4 dropped offline during the rebuild; this is typically a power/connection issue.

Well, the extended SMART test is still sitting at 10%. Since this is a Dell T30 I have to wrestle with the non-standard power situation. I've ordered a SFF PSU with ATX adapter plate hoping it will mount. Also ordered a 24-to-8 pin PSU adapter for motherboard power. The T30 was designed with space for 4 drives in it. I added a cage and 2 more drives. Maybe the PSU has gotten tired. Seems like a tired PSU would show other symptoms though.

Update: Extended SMART test didn't produce any errors. Clearning the drive again, then will try adding it again. New PSU ordered. That will leave the motherboard and HBA as possible problems.

Edited August 14, 2025Aug 14 by Abe677
Added update.

Solved (for now). JorgeB - thank you very much for your help.

I moved the drive to another SATA power connection, cleared the new drive again, the rebuilt the array (in maintenance mode). Array rebuilt without issue and the server is running.

The plan forward.

Because I had "shucked" a number of drives in this system, I used molex-to-SATA power splitters for those drives. I wanted to avoid taping pins. Seems like one of those splitters may have a problem. If I had a time machine I might go back and avoid shucked drives in this specific build. There are other configurations, like my Jonsbo N2, where shucked drives are fine and no splitters required.

Dell T30 server, so non-standard power inside. I have a SFF 750 watt PSU to replace the stock 290 watt unit. Waiting on a special 24-to-8 pin adapter for motherboard power (from the new PSU). When I next open the case I'll see how many potentional schucked drives I might have and make plans to replace them over time. The goal would be to get rid of the power splitters.