#3766 SMART status only shows first drive on 3ware RAID controller

1.520
closed-fixed
5
2010-10-07
2010-09-30
No

I am opening a new issue for this, as the others are already closed and somewhat old. On my Ubuntu 10.04 server with a 3ware 9550SX-8LP card, the SMART status page only shows the first disk on my RAID controller.

I believe I saw something in another report saying that the current logic attempts to read each 3ware port sequentially until it gets a failure, to determine how many drives are present. If this is the case, it is likely the cause of my problem; I may have something of an edge case here, but I have drives on ports 0, 3, 4, and 5.

Is there a better solution, besides re-arranging my drives? I'll do it if I have to, but I'm also very willing to assist in finding a better way.

Discussion

1 2 > >> (Page 1 of 2)
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    More recent versions of Webmin attempt to count the number of disks in a 3ware array using the command :

    tw_cli info c0

    What does that output on your system?

     
  • Steve McGrath

    Steve McGrath - 2010-09-30

    Actually for whatever reason, (maybe the slot it's installed in?) my controller is c2. Here is the output from tw_cli info c2:

    steve@cthulu:/dev$ sudo tw_cli info c2

    Unit UnitType Status %RCmpl %V/I/M Stripe Size(GB) Cache AVrfy
    ------------------------------------------------------------------------------
    u0 SINGLE OK - - - 232.82 OFF OFF
    u1 RAID-5 OK - - 64K 1862.62 OFF OFF

    Port Status Unit Size Blocks Serial
    ---------------------------------------------------------------
    p0 OK u0 232.88 GB 488397168 9VY94PFZ
    p1 NOT-PRESENT - - - -
    p2 NOT-PRESENT - - - -
    p3 OK u1 931.51 GB 1953525168 S246J1CZ700977
    p4 OK u1 931.51 GB 1953525168 S246J1CZ700987
    p5 OK u1 931.51 GB 1953525168 S246J1CZ700978
    p6 NOT-PRESENT - - - -
    p7 NOT-PRESENT - - - -

     
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    smart-status.wbm.gz

     
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    Thanks - I think I see the issue, which is that Webmin assumes that all the disks in the array are numbered 0,1,2,etc..

    Attached to this bug is an update to the Smart Status module that should fix the issue. You can install it at Webmin -> Webmin Configuration -> Webmin Modules.

     
  • Steve McGrath

    Steve McGrath - 2010-09-30

    Well, I installed the attached module, and it seems to make no difference. I still only see "3ware physical disk 0" in the drive selection dropdown.

     
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    Hmm .. what does it show as the /dev device name for the 3ware drive in the menu in the SMART module?

     
  • Steve McGrath

    Steve McGrath - 2010-09-30

    It actually doesn't show a device name in the menu, which is odd now that you mention it. If I proceed to select it and view the SMART info, it gives the device name as "/dev/twa0, disk 0"

     
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    So I guess you don't have c0 or c1 3ware devices?

     
  • Steve McGrath

    Steve McGrath - 2010-09-30

    Correct, there are no other RAID controllers in this server, and this one, for whatever reason, is c2. That's what the controller configuration shows as well, ID 2.

    Must be assigned based on PCI-X slot number, phase of the moon, or fair dice roll...

     
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    I think the drive being c2 is the issue, as Webmin expects the fitrst 3ware device to be c0.

    In your /etc/fstab file, what device name do you use for the mounts from the 3ware RAID array?

     
  • Nobody/Anonymous

    I'm using LVM, so my fstab has /dev/mapper devices listed rather than physical partitions.

    However, the devices visible to the operating system are /dev/sda and /dev/sdb.

    sda is 3ware port 0, a single drive for the operating system.
    sdb is 3ware ports 3, 4, 5 in a RAID-5 for storage.

    I will be taking this server down for maintenance later today, so while I'm at it I'll see if I can determine why it is being assigned ID 2.

    I note that the kernel sees this card as scsi2, with the onboard primary and secondary IDE being assigned scsi0 and scsi1, respectively. Perhaps this is why the controller is showing up as C2.

    It could still also be the slot number, or perhaps configurable in the controller's BIOS. I can check both during maintenance and experiment a bit.

    Thank you for looking in to this so quickly, Webmin is an amazing piece of software!

     
  • Jamie Cameron

    Jamie Cameron - 2010-09-30

    I'm pretty sure that if the 3ware device was on c0, Webmin would recognize it OK. Automatic detection that it is on c2 would be better though .. is there some 3ware command that can be used to discover that /dev/sda maps to c2 ?

     
  • Nobody/Anonymous

    I'm not aware of any specific command that would easily allow mapping /dev/sdX devices to their parent controller. But perhaps the output of "tw_cli show" could be parsed to determine what controller ids are present? Here's the output of that command on my server:

    Ctl Model (V)Ports Drives Units NotOpt RRate VRate BBU
    ------------------------------------------------------------------------
    c2 9550SX-8LP 8 4 2 0 1 1 -

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-02

    OK, thanks .. I think we are getting closer now.

    So I presume that u0 is /dev/sda , and u1 is /dev/sdb ?

     
  • Steve McGrath

    Steve McGrath - 2010-10-02

    Correct, at least in my case. These are the "units" that the are created in the 3ware BIOS, and made visible as disks to the host system.

    I've done some more pondering and researching, and I have some thoughts:

    I believe the 3ware driver calls this card c2 because it is scsi controller 2 in the system. It appears, from my googling, that the scsi controller id is assigned in the order that the kernel loads drivers. It probes IDE first, apparently, so the onboard IDE channels get scsi0 and scsi1, so when it loads the 3ware driver, it gets scsi2.

    As far as easily mapping /dev/sdX names to controller numbers, I'm not sure the best way. In my dmesg at least, I can see lines like these:

    [ 1.871715] sd 2:0:0:0: [sda] 488259584 512-byte logical blocks: (249 GB/232 GiB)
    [ 1.873636] sd 2:0:1:0: [sdb] 3906207744 512-byte logical blocks: (1.99 TB/1.81 TiB)

    Those lines show that sda is controller 2, channel 0, unit 0, LUN 0. I'm not sure how to retrieve that information any other way, though.

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-05

    Thanks for the info - I have one more question though. Does your system have /dev/twe* or /dev/twa* device files?

     
  • Steve McGrath

    Steve McGrath - 2010-10-05

    I have the /dev/twa* devices.

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-06

    smart-status.wbm.gz version 2

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-06

    Ok, I have attached a new module update that should help.

    If not, please let me know what lines get logged to /var/webmin/miniserv.error when you open the SMART module page.

     
  • Steve McGrath

    Steve McGrath - 2010-10-06

    I think we're getting closer now. The SMART module now lists the 4 physical disks in the drop down box. However, it reports that they do not support SMART, even after I manually enabled it and retrieved data with smartctl.

    I am attaching the tail end of my miniserv.error

     
  • Steve McGrath

    Steve McGrath - 2010-10-06

    miniserv.error after installing latest module

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-06

    smart-status.wbm.gz version 3

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-06

    Ok, thanks .. try out this even newer update.

    Also, what /dev/twa* files exist on your system?

     
  • Nobody/Anonymous

    I have /dev/twa[0-15].

    For some reason that last file wouldn't extract properly, I had to gunzip it on the command line, it extracted but with a "trailing garbage ignored" message. Once gunzipped, it installed, but still indicated no SMART support.

     
  • Jamie Cameron

    Jamie Cameron - 2010-10-06

    smart-status.wbm.gz version 4

     
1 2 > >> (Page 1 of 2)

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks