SSD errors - how to troubleshoot?

[tashirosgt]

I have an XFCE spin of Fedora Core 16 on a machine with kernel 3.3.4-1.fc16.x86_64 #1 SMP. It has a solid state drive on ata7.

May 2 20:06:03 comp02 kernel: [ 2.196312] ata7.00: ATA-8: INTEL SSDSA1NW080G3, 4PC10362, max UDMA/133
May 2 20:06:03 comp02 kernel: [ 2.196321] ata7.00: 156301488 sectors, multi 16: LBA48 NCQ (depth 0/32)
May 2 20:06:03 comp02 kernel: [ 2.199296] ata7.00: configured for UDMA/100

This is an installation from the LiveCd that was updated today. The SSD contains the "/" filesystem, except for /var, /tmp, swap and /home. Those filesystems are on laptop hard drive. I find the following kinds of errors in /var/log/messages:

May 3 17:40:09 comp02 kernel: [ 3607.331699] ata7.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0
May 3 17:40:09 comp02 kernel: [ 3607.331711] ata7.00: BMDMA2 stat 0x68651001
May 3 17:40:09 comp02 kernel: [ 3607.331720] ata7.00: failed command: DATA SET MANAGEMENT
May 3 17:40:09 comp02 kernel: [ 3607.331736] ata7.00: cmd 06/01:01:00:00:00/00:00:00:00:00/a0 tag 0 dma 512 out
May 3 17:40:09 comp02 kernel: [ 3607.331740] res 51/04:01:00:00:00/00:00:00:00:00/a0 Emask 0x1 (device error)
May 3 17:40:09 comp02 kernel: [ 3607.331748] ata7.00: status: { DRDY ERR }
May 3 17:40:09 comp02 kernel: [ 3607.331754] ata7.00: error: { ABRT }
May 3 17:40:09 comp02 kernel: [ 3607.331770] ata7.00: device reported invalid CHS sector 0
May 3 17:40:09 comp02 kernel: [ 3607.331784] sd 7:0:0:0: [sdf] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
May 3 17:40:09 comp02 kernel: [ 3607.331793] sd 7:0:0:0: [sdf] Sense Key : Aborted Command [current] [descriptor]
May 3 17:40:09 comp02 kernel: [ 3607.331803] Descriptor sense data with sense descriptors (in hex):
May 3 17:40:09 comp02 kernel: [ 3607.331808] 72 0b 00 00 00 00 00 0c 00 0a 80 00 00 00 00 00
May 3 17:40:09 comp02 kernel: [ 3607.331825] 00 00 00 00
May 3 17:40:09 comp02 kernel: [ 3607.331834] sd 7:0:0:0: [sdf] Add. Sense: No additional sense information
May 3 17:40:09 comp02 kernel: [ 3607.331844] sd 7:0:0:0: [sdf] CDB: Write same(16): 93 08 00 00 00 00 00 3d c6 e8 00 00 00 08 00 00
May 3 17:40:09 comp02 kernel: [ 3607.331867] end_request: I/O error, dev sdf, sector 4048616

Sometimes a yum update or install is apparently interrupted by these errors. They are printed on the terminal running yum and I end up with "uncompleted transactions", i.e. the problem appears to happen after the downloads of the rpms are finished and they are being installed.

What's a good way to troubleshoot this problem?

The machine is buggy. It occasionally has a "hard" freeze, meaning the cursor doesn't show and numlock doesn't work.

Edit: A screwy thing about my /var/log/messages is that the messages don't appear in chronological order.. Is this a known bug?

May 2 20:09:54 comp02 dbus-daemon[1020]: dbus[1020]: [system] Successfully activated service 'org.freedesktop.RealtimeKit1'
May 3 02:09:54 comp02 rtkit-daemon[1350]: Successfully made thread 1349 of process 1349 (/usr/bin/pulseaudio) owned by '42' high priority at nice level -11.
May 3 02:09:55 comp02 rtkit-daemon[1350]: Successfully made thread 1354 of process 1349 (/usr/bin/pulseaudio) owned by '42' RT at priority 5.
May 3 02:09:55 comp02 rtkit-daemon[1350]: Successfully made thread 1355 of process 1349 (/usr/bin/pulseaudio) owned by '42' RT at priority 5.
May 2 20:09:55 comp02 gnome-session[1287]: DEBUG(+): GsmManager: RegisterClient 103b13caf88174f0e4133601099342950800000012870001
May 2 20:09:55 comp02 gnome-session[1287]: DEBUG(+): GsmManager: Adding new client 103b13caf88174f0e4133601099342950800000012870001 to session

Edit: Answer, yes it is a known bug with rtkit-daemon: https://bugzilla.redhat.com/show_bug.cgi?id=767314

[JEO]

Found a bugzilla where others are having similar problems with an SATA SSD F15 through F17. Doesn't seem to be resolved though.

https://bugzilla.redhat.com/show_bug.cgi?id=680667

[tashirosgt]

Thank you for that link! The error messages in the logs look just like mine

I'll try some random things. (Isn't that the usual procedure for fixing computer problems?) I can set the SATA controller to IDE instead of AHCI. I also have another system with an SSD (different MB, same model SSD) that I'll be installing FC 16 upon.


.

[stevea]

Is there a different sata controller on the mobo to try ?

[george_toolan]

A SSD should be connected to the first port of your motherboard's internal controller which usually only has six ports.

Why did you connect it to an external controller?

What other "optimizations" did you try?

[tashirosgt]

Originally Posted by stevea

Is there a different sata controller on the mobo to try ?

Yes. (GA-K8N Ultra SLI). I'll try it.

This trial and error would be simpler if there was a diagnostic program for SSDs. I'll have to see if Intel has one.

Edit: There is http://downloadcenter.intel.com/Deta...dID=18455#help, which is only for every flavor of Windows.

Is this another opportunity for me to blacklist Intel SSDs or will all manufacturers of SSDs only have Windows diagnostic programs?

Running a Windows diagnostic program for SSDs with wine is an idea, but not one I'm eager to try.

Edit: There's smartctl. It has been installed (either in the original install or in something I added to it). I've never found the SMART technology helpful but my Intel SSD does support it.

[george_toolan]

Also try a different sata cable.

If you read the description of this diagnostic program you will realize you don't need this at all, because you already have smartctl, hdparm and fstrim.

smartctl -a /dev/sdf

[tashirosgt]

Originally Posted by george_toolan

If you read the description of this diagnostic program

If you read it, it says it can do diagnostic scans. Are you saying any of the Linux utilities can do them?

[george_toolan]

smartctl -t short /dev/sdf

SMART Self-test log structure revision number 1
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Vendor (0x98)       Completed without error       00%      1256         -

[tashirosgt]

In attempting to determine whether the error messages deal with reading or writing to the SSD, I find that the error:

May 3 17:40:09 comp02 kernel: [ 3607.331699] ata7.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0
May 3 17:40:09 comp02 kernel: [ 3607.331711] ata7.00: BMDMA2 stat 0x68651001
May 3 17:40:09 comp02 kernel: [ 3607.331720] ata7.00: failed command: DATA SET MANAGEMENT
May 3 17:40:09 comp02 kernel: [ 3607.331736] ata7.00: cmd 06/01:01:00:00:00/00:00:00:00:00/a0 tag 0 dma 512 out
May 3 17:40:09 comp02 kernel: [ 3607.331740] res 51/04:01:00:00:00/00:00:00:00:00/a0 Emask 0x1 (device error)

is probably related to TRIM. So says:http://en.wikipedia.org/wiki/TRIM

ATA

The TRIM command specification[46] is being standardized as part of the AT Attachment (ATA) interface standard, led by Technical Committee T13 of the International Committee for Information Technology Standards (INCITS).[47] TRIM is implemented under the DATA SET MANAGEMENT command (opcode 06h) the draft ACS-2 specification.[48]

[george_toolan]

Did you enable trim aka discard?

If you did then try to disable it and see if the errors disappear.

[tashirosgt]

Originally Posted by george_toolan

Did you enable trim aka discard?

If you did then try to disable it and see if the errors disappear.

I did enable it and I'll get around to disabling it. My next step has been to run updates on another system that I'm building. It has the same model SSD on a different motherboard. This second system has the discard option set and there aren't any ata DATA SET MANAGEMENT errors in the logs.