Is my HBA dying? (HP H240)

[EasyRhino]

Hello!

A few months ago I picked up a used HP H240 controller. It's running the current 7.0 firmware in non-raid mode.

lately I've noticed these errors in the master console of my debian VM I have it passed thru to:

blk_update_request: I/O error, dev sdb, sector 5560583976 op 0x0:(READ) flags 0x80700 phys_seg 14 prio class 0
[ 2033.450551] sd 3:0:1:0: [sdb] tag#134 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE cmd_age=2s
[ 2033.450569] sd 3:0:1:0: [sdb] tag#134 Sense Key : Aborted Command [current] [descriptor]
[ 2033.450573] sd 3:0:1:0: [sdb] tag#134 Add. Sense: Information unit iuCRC error detected
[ 2033.450576] sd 3:0:1:0: [sdb] tag#134 CDB: Read(16) 88 00 00 00 00 01 4b 6f c7 28 00 00 00 80 00 00

I saw them a LOT with a SAS3 SSD (a Netapp X371 / Samsung PM1635a), but also see them on a HGST SAS3 spinner HDD.

I tried two different SAS breakout cables and both had the errors.

the other card I have is an old LSI 9260-4i, which is hardware RAID and SAS2, so it's a different beastie. But I don't get the linux console errors with it.

I can just replace whatever is broken but i'm not positive where to start. I've got a weird SSD paired with a weird card with weird cabling.

And if it is the controller I wonder if anyone else has seen these problems with H240s?

 

[EasyRhino]

Well I replaced the h240 with another h240 and identical problems. Lots of problems with the sas SSD and some errors with a sata HDD.

worse, when I was going through multiple passes of sg_format on the SSD, one got interrupted and I think I bricked my SSD. Now it doesn't appear on the list of drives at all with the H240. And my old LSI 9260 shows it as "unconfigured (bad)" and throws an error when I try to change it to unconfigured (good).

 

[ecosse]

I would tick liked but it seems inappropriate - I feel your pain. I/O error did sound like the drive but these links suggests the things you suspected are at fault.

Solved - ZFS faulted | The FreeBSD Forums
Extended Sense Data from Device - Hewlett Packard Enterprise Community (hpe.com)

May turn out to be something else but I have four of those h240's and every time I am tempted to use one I read something like the above. I don't trust them.

 

[EasyRhino]

Yes well the latest update is that the SSD is now dead as a doornail.

while trying to debug, for a while it would be 'half detected' by the controller. Like it was showing the model number, but no info on capacity. and plugging into my old LSI 9260 hardware raid would show the drive unconfigured (Bad) and it would error if trying to change it to good.

But now, after sitting around for a few hours plugged in and several hours unplugged, the drive is super dead, and not recognized at all.

still unsure if what I saw was a gradual failure of the drive itself, or if it was self-inflicted by my attempts to mess around with it.

 

[ecosse]

You could try seatools or the like. Some of those lower level tools might be able to resurrect. I would change the cabling if you havent already. Unconfigured (bad) doesn't necessarily mean its a bad drive, it means it was unexpectedly powered off or disconnected.