[SATA Controller Cards]

On 7/30/2019 at 7:07 PM, JorgeB said:

This one should work with the standard AHCI driver, though usually don't work very well because of the port multipliers.

 

That is indeed the case. Here's my story of trying the

 

"PCIe 2.0 X4 to SATA III 6-Port Adapter Card (ASM Chipset) for IPFS Mining - Model SU-SA3006.V1"

 

card in my new server. It first seemed to "work" as far as it was capable of detecting and reading from disks. However, as soon as I tried to write to the array, it started producing an ever more frequent series of errors which made me believe that I had bad disks. This is just a small excerpt of the logs:

 

Jan 15 02:28:33 pstore kernel: ata7.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
Jan 15 02:28:33 pstore kernel: ata7.00: failed command: FLUSH CACHE EXT
Jan 15 02:28:33 pstore kernel: ata7.00: cmd ea/00:00:00:00:00/00:00:00:00:00/a0 tag 18
Jan 15 02:28:33 pstore kernel:         res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
Jan 15 02:28:33 pstore kernel: ata7.00: status: { DRDY }
Jan 15 02:28:33 pstore kernel: ata7: hard resetting link
Jan 15 02:28:34 pstore kernel: ata7: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan 15 02:28:39 pstore kernel: ata7.00: qc timeout (cmd 0x27)
Jan 15 02:28:39 pstore kernel: ata7.00: failed to read native max address (err_mask=0x4)
Jan 15 02:28:39 pstore kernel: ata7.00: HPA support seems broken, skipping HPA handling
Jan 15 02:28:39 pstore kernel: ata7.00: revalidation failed (errno=-5)
Jan 15 02:28:39 pstore kernel: ata7: hard resetting link
Jan 15 02:28:39 pstore kernel: ata7: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan 15 02:28:45 pstore kernel: ata7.00: qc timeout (cmd 0xef)
Jan 15 02:28:45 pstore kernel: ata7.00: failed to enable AA (error_mask=0x4)
Jan 15 02:28:45 pstore kernel: ata7.00: revalidation failed (errno=-5)
Jan 15 02:28:45 pstore kernel: ata7: limiting SATA link speed to 3.0 Gbps
Jan 15 02:28:45 pstore kernel: ata7: hard resetting link
Jan 15 02:28:45 pstore kernel: ata7: SATA link up 6.0 Gbps (SStatus 133 SControl 320)
Jan 15 02:29:00 pstore kernel: ata7.00: qc timeout (cmd 0xef)
Jan 15 02:29:00 pstore kernel: ata7.00: failed to set xfermode (err_mask=0x4)
Jan 15 02:29:00 pstore kernel: ata7.00: disabled
Jan 15 02:29:01 pstore kernel: ata7: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
Jan 15 02:29:01 pstore kernel: ata7: EH complete
Jan 15 02:29:01 pstore kernel: sd 7:0:0:0: [sdb] tag#20 UNKNOWN(0x2003) Result: hostbyte=0x04 driverbyte=0x00
Jan 15 02:29:01 pstore kernel: sd 7:0:0:0: [sdb] tag#20 CDB: opcode=0x88 88 00 00 00 00 00 00 00 00 40 00 00 00 08 00 00
Jan 15 02:29:01 pstore kernel: print_req_error: I/O error, dev sdb, sector 64
Jan 15 02:29:01 pstore kernel: md: disk1 read error, sector=0
Jan 15 02:29:01 pstore kernel: sd 7:0:0:0: [sdb] tag#22 UNKNOWN(0x2003) Result: hostbyte=0x04 driverbyte=0x00
Jan 15 02:29:01 pstore kernel: sd 7:0:0:0: [sdb] tag#22 CDB: opcode=0x35 35 00 00 00 00 00 00 00 00 00
Jan 15 02:29:01 pstore kernel: print_req_error: I/O error, dev sdb, sector 2930303272
Jan 15 02:29:01 pstore kernel: md: disk1 write error, sector=2930303208
Jan 15 02:29:01 pstore kernel: print_req_error: I/O error, dev sdb, sector 2930303264

 

After replacing all cables and two drives to no avail, I remembered that I had another cheap multiport SATA card lying around, and lo and behold: no more errors! The present card is a PCE8SAT-M01 (around $20 from Aliexpress) which I thought would be even crappier than the 6-port. It only has one PCIe lane so I wasn't surprised that it only transfers around 430 MB/sec (six "old"/PMR WD Red drives) - but at least it does that without conking out!

 

 

If you're interested in some technical background, I found an excellent article on its performance:

 

8 Port SATA on a PCIE 1x lane

 

While I'm waiting for a card with a bit more "oomph", this cheap controller is serving me well to keep the server running until then.