Extremely Slow Read/Writes with SSDs

[rcurrier8]

I am trying my hand at creating my first server for media. I am just messing around with the hardware using spare SSDs prior to buying all new hard drives to learn the software. I am seeing incredibly slow read/writes when doing the parity check and when using preclear disk. I thought this may be a CPU limitation, so I purchased an upgrade with no change. Then thought my HBA may be the culprit, updated it and still no change. Ran “Fix Common Problems” and no errors were found.

Am I wrong with thinking this is too slow? With (IMO) good hardware and SSDs I was expecting read/writes of at least 100 mb/s.

Current hardware:

MB: Intel S2600CP (12/23/13 BIOS as Intel no longer supports MBs so no update is available)

CPU: Xeon E5-2697 V2 (x1 but have the option to run dual CPUs)

Ram: 128GB ECC DDR3

HBA: LSI 9211-8i updated to V20 in IT mode

Hard Drives: 6x Samsung 840 1tb, 3x Intel 520 series 480GB (one to be used as cache)

Attached is the diag files as well as the disk smart files.

 

 

tower-diagnostics-20220126-1346.zip DISK_SMART_FILES.zip

[JorgeB]

The 840 EVO can turn slow reading old data, especially if not using the latest firmware, run the diskspeed docker to test read performance in all devices.

[Squid]

sdd (disk 1) seems to have connection problems and is suffering from device resets slowing everything down.  Reseat the cabling

[rcurrier8]

59 minutes ago, Squid said:

sdd (disk 1) seems to have connection problems and is suffering from device resets slowing everything down.  Reseat the cabling

I removed this disk, started a new array and the speeds have significantly improved. For my knowledge how could you see that the disk was having connection problems (where in the logs)?

[Squid]

Jan 26 10:25:10 Tower kernel: sd 2:0:2:0: attempting task abort!scmd(0x00000000e31eeef1), outstanding for 30382 ms & timeout 30000 ms
Jan 26 10:25:10 Tower kernel: sd 2:0:2:0: [sdd] tag#3057 CDB: opcode=0x2a 2a 08 3a 38 3e a8 00 00 08 00
Jan 26 10:25:10 Tower kernel: scsi target2:0:2: handle(0x000a), sas_address(0x4433221101000000), phy(1)
Jan 26 10:25:10 Tower kernel: scsi target2:0:2: enclosure logical id(0x5000000080000000), slot(2) 
Jan 26 10:25:11 Tower kernel: sd 2:0:2:0: task abort: SUCCESS scmd(0x00000000e31eeef1)
Jan 26 10:25:11 Tower kernel: sd 2:0:2:0: Power-on or device reset occurred
### [PREVIOUS LINE REPEATED 1 TIMES] ###
Jan 26 10:25:13 Tower rc.diskinfo[8139]: SIGHUP received, forcing refresh of disks info.
Jan 26 10:25:46 Tower kernel: sd 2:0:2:0: attempting task abort!scmd(0x0000000038bf96d8), outstanding for 30329 ms & timeout 30000 ms
Jan 26 10:25:46 Tower kernel: sd 2:0:2:0: [sdd] tag#3072 CDB: opcode=0x2a 2a 08 3a 38 3e a8 00 00 08 00
Jan 26 10:25:46 Tower kernel: scsi target2:0:2: handle(0x000a), sas_address(0x4433221101000000), phy(1)
Jan 26 10:25:46 Tower kernel: scsi target2:0:2: enclosure logical id(0x5000000080000000), slot(2) 
Jan 26 10:25:47 Tower kernel: sd 2:0:2:0: task abort: SUCCESS scmd(0x0000000038bf96d8)
Jan 26 10:25:47 Tower kernel: sd 2:0:2:0: Power-on or device reset occurred
### [PREVIOUS LINE REPEATED 1 TIMES] ###
Jan 26 10:25:50 Tower rc.diskinfo[8139]: SIGHUP received, forcing refresh of disks info.
### [PREVIOUS LINE REPEATED 1 TIMES] ###
Jan 26 10:27:21 Tower kernel: sd 2:0:2:0: attempting task abort!scmd(0x0000000074c99a4d), outstanding for 30139 ms & timeout 30000 ms
Jan 26 10:27:21 Tower kernel: sd 2:0:2:0: [sdd] tag#3107 CDB: opcode=0x2a 2a 08 3a 38 3e b0 00 00 08 00
Jan 26 10:27:21 Tower kernel: scsi target2:0:2: handle(0x000a), sas_address(0x4433221101000000), phy(1)
Jan 26 10:27:21 Tower kernel: scsi target2:0:2: enclosure logical id(0x5000000080000000), slot(2) 
Jan 26 10:27:21 Tower kernel: sd 2:0:2:0: task abort: SUCCESS scmd(0x0000000074c99a4d)
Jan 26 10:27:21 Tower kernel: sd 2:0:2:0: Power-on or device reset occurred
Jan 26 10:27:22 Tower rc.diskinfo[8139]: SIGHUP received, forcing refresh of disks info.
Jan 26 10:27:22 Tower kernel: sd 2:0:2:0: Power-on or device reset occurred