Running smartctl from shell causing hba timeouts

[Turgin]

I have 8 x 2TB SATA2 drives attached to an IBM BR10i (LSI 1068e). I also have 6 x 128GB SSDs attached to the LSI controller on my Supermicro X8DT6-F motherboard. Both HBAs are flashed with IT firmware.

New install of FreeNAS-9.10-STABLE-201605240427 (64fcd8e).

I am playing with running FreeNAS under ESXi 6 for an AIO home server. My FreeNAS guest has 32GB RAM (locked), 2 vCPU, and I have PCI passthrough configured for both HBAs. Everything seems to be working well except..

When I run smartctl -a on any of the 2TB drives from the shell option through the ESXi console I get the following errors:

Code:

mpt1: request 0xfffffe00009e1da8:2550 timed out for ccb 0xfffff8000b794800 (req->ccb 0xfffff8000b794800)
mpt1: attempting to abort req 0xfffffe00009e1da8:2550 function 0
mpt1: completing timedout/aborted req 0xfffffe00009e1da8:2550
mpt1: abort of req 0xfffffe00009e1da8:0 completed
mpt1: request 0xfffffe00009f3d08:2552 timed out for ccb 0xfffff8000b794800 (req->ccb 0xfffff8000b794800)
mpt1: attempting to abort req 0xfffffe00009f3d08:2552 function 0
mpt1: completing timedout/aborted req 0xfffffe00009f3d08:2552
mpt1: abort of req 0xfffffe00009f3d08:0 completed

I do not get them when running short or long tests nor do I get these errors when running the same commands from an ssh session (nothing in dmesg either) . I also do not get these errors on the SSDs from either console.

Should I be concerned? Any ideas why its happening?

 

[Mirfster]

IBM BR10i (LSI 1068e)

Not sure, but from what I gather this is a pretty old card and cannot see drive space over 2.2 TB (like the PERC 6/I). May want to consider getting a more commonly recommended card (M1015, LSI 9211-8I, H200, etc..) and possibly kill two birds with one stone.

If your hearts set on sticking with those cards, then perhaps others will provide some insight.

 

[Turgin]

Not sure, but from what I gather this is a pretty old card and cannot see drive space over 2.2 TB (like the PERC 6/I). May want to consider getting a more commonly recommended card (M1015, LSI 9211-8I, H200, etc..) and possibly kill two birds with one stone.

If your hearts set on sticking with those cards, then perhaps others will provide some insight.

I'm not set on it necessarily , but I'm only using 2TB SATA2 drives on it so I thought it would be ok to use for now. The drives are not recognized by the Supermicro onboard LSI (but are by the onboard SATA) and I tried this so I could get 8 drives without having to get another card.

I fully intend to get an M1015 when I change out the drives for 4TB ones down the road.

 

[cyberjock]

If memory serves me right something isn't quite right with those cards when doing PCIe passthrough.

Even if I'm wrong, the best fix I see is to get an M1015 or equivalent.

 

[Turgin]

If memory serves me right something isn't quite right with those cards when doing PCIe passthrough.

Even if I'm wrong, the best fix I see is to get an M1015 or equivalent.

I was afraid of something like that.

Unfortunately, these 2TB drives aren't recognized by the onboard LSI 2008 and I suspect an add-on card with the same chipset would yield the same result. The drives were working pulls from a decommissioned NetApp filer and they have some "special" NetApp firmware which apparently interferes with them working on SAS2/SATA3 controllers for some reason. I even contacted Hitachi trying to get some updated firmware and they referred me to NetApp based on the serial number. I didn't pursue that any further.

For the moment, since this is still a sandbox for me, I'll just roll the dice unless someone can recommend a SATA2 HBA not based on the 1068e chipset to try.

 

[Mirfster]

Well, you can get a H200 for ~$55.00 (without cables) if you are so inclined. Not sure if it is going to cure your drives since you mentioned that they came from a NetApp. While I am not really versed in NetApp drives, I feel a bit leery about them especially if Hitachi pass you off to them... But would think that maybe/perhaps there is a way to wipe them...

 

[Turgin]

But would think that maybe/perhaps there is a way to wipe them...

One would think...I couldn't convince the Hitachi support guy to give me any of the secret sauce. I tried the firmware for a "normal" DK7SAA200 drive and the flash utility wouldn't do it because it didn't like the firmware on the drive already. I may be trying too hard to make something work that is probably going to make me cry in the future.

 

[Mirfster]

Well, according to this thread it seems like a dead end: Netapp Branded Hitachi SAS Drives - Any workaround?
But this thread may give some hope: Poor performance with certain SAS drives... lets tag @tvsjr and see what they may mention about drives that have spent some time in a NetApp...

 

[cyberjock]

Netapp stuff is basicallt proprietary and locked down. So yeah, you're really making a mistake trying to use anything that is Netapp hardware in anything that isn't a Netapp device. This issue is actually a selling point when people call iXsystems for a quote. ;)

 

[Turgin]

Netapp stuff is basicallt proprietary and locked down. So yeah, you're really making a mistake trying to use anything that is Netapp hardware in anything that isn't a Netapp device. This issue is actually a selling point when people call iXsystems for a quote. ;)

Fair enough. Since I am still testing and playing around they seem to work well enough. I won't trust them for any "real" data once I get to that point.