Yorick
Wizard
- Joined
- Nov 4, 2018
- Messages
- 1,912
For testing purposes, I bought an IBM M1015 "already in IT mode! FreeNAS UnRaid come buy now!" and a lot of six (6) Seagate 2TB SAS drives off eBay.
Today I got around to plugging all that in, and thought I'd share my results. It might be helpful to someone else.
Three of the drives would start badblocks, the other three would not. "invalid starting block (0): must be less than 0". Well that made no sense.
Edit: That may have been an artifact of using /dev/sgX. All drives give these error message on sgX. sdX works with badblocks. That said, the 520-byte issue below was real, and badblocks wouldn't work until it was corrected.
I have a mixed lot: Three ST32000444SS drives, and three ST2000NMCLAR2000. It was the latter that refused to do badblocks, wouldn't show up in fdisk or parted, and couldn't be added to a test pool.
It turns out they were formatted with 520 byte blocks, likely came out of an EMC array. Converting them to 512 bytes should solve that, I'm doing that now. Instructions: https://mikeyurick.com/reformat-emc...tems-520-to-512-block-size-conversion-solved/
The HBA was on version 10.00.08 and showed the Product ID as "Undefined" and the Controller as "SAS2008(??)". The Vendor was "LSI", so the eBay seller at least got it partway cross-flashed, just stopped shy of being done. I flashed it to P20, and now it's version 20.00.07, Product ID SAS9211-8i, and Controller SAS2008(B2). Better.
Question: I know Chinese knock-offs are being sold on eBay.How would I identify one of those? If it takes an LSI firmware, is it an LSI device? It shows up as as "Board Name" "IBM 6Gb perf HBA" after flash to P20. I'm assuming it's legit; it'd be good to know for sure.
Edit: Not a knockoff after all, it's likely a sweatshop pull.
This is the one I have: https://www.ebay.com/itm/LSI-IBM-SA...S-unRAID-6Gbps-SAS-HBA-US-seller/132748694996
It's the IBM 6Gb SAS Performance optimized HBA , as a server pull at https://www.ebay.com/itm/46M0912-IB...-PCIe-x8-LSI-SAS9200-8i-46C8937-/382957256585 .
@Stilez guide to cross-flashing at https://www.ixsystems.com/community...-lsi-9211-9300-9305-9311-hba-and-variants.54/ was easy to follow and gave me everything I needed in one convenient package. Thank you.
Getting into EFI wasn't too difficult, it just wasn't possible by choosing the USB stick as a UEFI boot device. Instead I went into UEFI Setup, Advanced, then Exit, and chose "launch EFI shell". That's on an old ASUS board.
"Grown defect list" on the three ST32000444SS drives is 25, 53 and 180 - I don't expect them to last very long. Which is okay. I wanted them for testing, not live data, and they were dirt cheap. The ST2000NMCLAR2000 have a grown defect list of 0 so far, still formatting the third one.
I'll update once re-format and then badblocks is complete, with a note on whether they became usable (I expect yes) and how badly they were damaged.
Edit:Oh goodness I think I might have a knockoff, unless there were different versions of these around.
And this is one from eGoods. Note slightly different silk screening: https://www.ebay.com/itm/IBM-ServeRaid-M1015-46C8933-SAS-SATA-PCI-e-RAID-CONTROLLER/282202739354
Recommendations? Can I prove with some certainty that this is or isn't a knockoff? If it is, I should likely get a refund through eBay and buy a replacement.
Today I got around to plugging all that in, and thought I'd share my results. It might be helpful to someone else.
Three of the drives would start badblocks, the other three would not. "invalid starting block (0): must be less than 0". Well that made no sense.
Edit: That may have been an artifact of using /dev/sgX. All drives give these error message on sgX. sdX works with badblocks. That said, the 520-byte issue below was real, and badblocks wouldn't work until it was corrected.
I have a mixed lot: Three ST32000444SS drives, and three ST2000NMCLAR2000. It was the latter that refused to do badblocks, wouldn't show up in fdisk or parted, and couldn't be added to a test pool.
It turns out they were formatted with 520 byte blocks, likely came out of an EMC array. Converting them to 512 bytes should solve that, I'm doing that now. Instructions: https://mikeyurick.com/reformat-emc...tems-520-to-512-block-size-conversion-solved/
The HBA was on version 10.00.08 and showed the Product ID as "Undefined" and the Controller as "SAS2008(??)". The Vendor was "LSI", so the eBay seller at least got it partway cross-flashed, just stopped shy of being done. I flashed it to P20, and now it's version 20.00.07, Product ID SAS9211-8i, and Controller SAS2008(B2). Better.
Question: I know Chinese knock-offs are being sold on eBay.
Edit: Not a knockoff after all, it's likely a sweatshop pull.
This is the one I have: https://www.ebay.com/itm/LSI-IBM-SA...S-unRAID-6Gbps-SAS-HBA-US-seller/132748694996
It's the IBM 6Gb SAS Performance optimized HBA , as a server pull at https://www.ebay.com/itm/46M0912-IB...-PCIe-x8-LSI-SAS9200-8i-46C8937-/382957256585 .
@Stilez guide to cross-flashing at https://www.ixsystems.com/community...-lsi-9211-9300-9305-9311-hba-and-variants.54/ was easy to follow and gave me everything I needed in one convenient package. Thank you.
Getting into EFI wasn't too difficult, it just wasn't possible by choosing the USB stick as a UEFI boot device. Instead I went into UEFI Setup, Advanced, then Exit, and chose "launch EFI shell". That's on an old ASUS board.
"Grown defect list" on the three ST32000444SS drives is 25, 53 and 180 - I don't expect them to last very long. Which is okay. I wanted them for testing, not live data, and they were dirt cheap. The ST2000NMCLAR2000 have a grown defect list of 0 so far, still formatting the third one.
I'll update once re-format and then badblocks is complete, with a note on whether they became usable (I expect yes) and how badly they were damaged.
Edit:
And this is one from eGoods. Note slightly different silk screening: https://www.ebay.com/itm/IBM-ServeRaid-M1015-46C8933-SAS-SATA-PCI-e-RAID-CONTROLLER/282202739354
Recommendations? Can I prove with some certainty that this is or isn't a knockoff? If it is, I should likely get a refund through eBay and buy a replacement.
Last edited: