I have an operational AMD EPYC based TrueNas Scale server using eight 6TB SAS drives. The server is just a test environment, so nothing mission critical and can be blown away without consequence. I wanted to get a few more 6TB SAS drives to try out additional storage configurations and bought what I thought were identical 6TB SAS drives to what I currently had (HGST). However, when the drives showed up, they were HPE smkr60000S5xen7.2 drives.
I attached the drives and while they are detected, they show as 0 bytes and I can’t initialize them. I also tried booting the server into Windows and it too can see the drives, but they cant be initialized. I looked up the drives online, and it appears they were originally intended to be used in a 3PAR storage solution. This makes me think its probably a firmware issue and perhaps due to block sizing 512e/4096/etc.
Does anyone have any idea on what I could do to make these drives work? Or am I hosed and have to look into returning these or potentially eating the cost of the drives?
But I had a similar issue the other day working with some dual actuator sas drives. I needed to mask off the 3v pins. But I didn’t do a great job. And one of the drives was “detected” meaning I could see it, and the two lungs, but I couldn’t see a size or do anything with it. Hours later, I decided to retry taping off the pins, and that solved it.
(Except for the part where truenas scale doesn’t like dual actuator drives much. But that’s a different problem).
The kapton tape mod didnt work. It actually made the drive not spin up (made it worse) on the one I tried it on. Having said this, I booted the machine up on an Ubutunu live CD and am attempting to change the drives block size via sg_format. None of the drives have finished the format yet, so still dont know if it will work or not but I will report back. Drives have been formatting for 12 hours thus far and are 78% complete (6TB 7200rpm).
EDIT - UPDATE:
Using sg_format in Ubuntu fixed the block size! The drives work in Windows and TrueNas now. This is awesome! Ironically, I ran across this post from Wendell explaining the exact issue and fix. Not sure why it didnt show up when I was searching the site, but here it is in case anyone else stumbles across this post: