• Bug#266200: [smartmontools-support]Bug#266200: smartctl.8.gz: doc tips

    From Bruce Allen@1:229/2 to Dan Jacobson on Sat Aug 21 01:30:15 2004
    From: [email protected]

    On Fri, 20 Aug 2004, Dan Jacobson wrote:

    I would put in a -s flag with a self-test schedule too.

    OK, note however that the -s examples are all for machines
    continuously running. For those computers are powered up for
    irregular lengths of time at irregular intervals, you need to add some anacron(8) examples or examples like how our disks are fdisk(8)
    checked at boot every 26 boots.

    The -s Directive will work for disks that are powered up for irregular
    lengths of time (eg, laptops) however no test will run at a scheduled time unless the machine is actually running then. On my laptop I schedule self-tests during times of the week when the machine is often up and
    running.

    BTW, it doesn't mention if one can use more than one -s, or must mash
    them together.

    Try more than one -s, and you'll get a warning message that all but the
    last are ignored.

    Wait, smartctl -a reveals that this is a "recommended time", therefore
    the message we see telling us to wait should say "at least" or
    "probably", etc.

    Message says it will complete AFTER stated time: this is correct.

    BTW, it would be nice if the date when that test was made would come
    out in that output too. One hopes that that date was stored somewhere
    on this mysterious drive filesystem.

    Only thing stored is power-on hours.

    P.S. "offline" is not explained in the docs.

    It's explained in the smartctl man page. See the -o option.

    Cheers,
    Bruce



    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)
  • From Dan Jacobson@1:229/2 to All on Fri Aug 20 21:00:26 2004
    From: [email protected]

    I would put in a -s flag with a self-test schedule too.

    OK, note however that the -s examples are all for machines
    continuously running. For those computers are powered up for
    irregular lengths of time at irregular intervals, you need to add some anacron(8) examples or examples like how our disks are fdisk(8)
    checked at boot every 26 boots.

    Furthermore, the -s discussion just have various examples, and no
    advice for "what is right for me".

    It should say what the light, medium, and heavy disk user might
    prefer, and the pros and cons and costs of each type of test.

    I'll just wildly pick the first example. -s S/../.././02

    I mean we all know we are supposed to see the dentist every six months
    for a checkup, so how about some similar clear tips for each type of user.

    BTW, it doesn't mention if one can use more than one -s, or must mash
    them together.

    If you look at the output of 'smartctl -v'

    Say I'll try it,
    smartctl --help
    smartctl -v /dev/hdb
    smartctl -v 9,temp /dev/hdb
    smartctl -h
    I suppose my disk doesn't support temperature. OK, never mind.

    I've had tests take more than 24 hours to run on busy machines.

    I bet it still just said to wait 31 minutes though.

    Wait, smartctl -a reveals that this is a "recommended time", therefore
    the message we see telling us to wait should say "at least" or
    "probably", etc.

    hunt down and either remap

    Well, that is changing the disk. So they are more than mere
    "tests". Sure hope the man page drives this point home.

    The self-test log is not stored in a file, it is stored in the drive.

    The man page should mention that.

    You run 'smartctl -l self-test /dev/hda' to extract the log from the

    No dash. The man page often is renderd with a dash there too.

    BTW, it would be nice if the date when that test was made would come
    out in that output too. One hopes that that date was stored somewhere
    on this mysterious drive filesystem.

    smartd will poll the self-test log and email someone if new failed

    Only if you tell it to email, apparently.

    I wanted to monitor the progress of the check but didn't know how.

    watch -d 'smartctl -c /dev/hda 2>&1 | grep ^Self-test.execution.status'

    Add that to the manpage.

    P.S. "offline" is not explained in the docs.

    Also we read -v modifies things. Maybe it even modifies the disk itself?


    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)