Problem with Hal on 2007.1 - Mandriva

This is a discussion on Problem with Hal on 2007.1 - Mandriva ; Hal is taking 97-99% of cpu and stalls the bootup process for several minutes! After upgrading a 2007 to 2007.1 system, I had a problem with stalled boot process going into any multiuser runlevel. So I decided to do a ...

+ Reply to Thread
Results 1 to 6 of 6

Thread: Problem with Hal on 2007.1

  1. Problem with Hal on 2007.1

    Hal is taking 97-99% of cpu and stalls the bootup process for several minutes!

    After upgrading a 2007 to 2007.1 system, I had a problem with stalled boot process going into any multiuser runlevel.

    So I decided to do a clean installation, reformatting every partition but /home and /usr/local. I used for the new ftp install the
    boot.iso image I discovered in an "official" 2007.1 mandriva
    repository--this was the same disk I used for the upgrade from 2007.0.

    The install went well enough (didn't detect monitor) until booting up the new installation for the first time. It appeared to hang up during service initialization at about the exact same place as the "upgraded" 2007->2007.1 installation would hang. Starting up with vga=normal so I can read all messages, I would see "Starting alsa v.10.12:intel8x0 alsactl: save.state: 1253 no sound cards found
    can't open /root/tmp/asound.state.tmp setting mixer settings alsactl: load_state: no sound cards found [failed]

    then it moves on to Hal daemon startup. HAL startup sits there literally for a few MINUTES before "fail" appears and the boot process
    finally continues on with other services. The first time I saw that I thought it would never complete booting into a runlevel, and I'd just have to wipe the partitions and start over with a different distro (which I may yet have to do). Probably this was exactly the same problem as with the upgraded 2007->2007.1 installation, but in that previous case I decided that the system was never going to finish booting up or was crashed, and made the decision to hit the reset button and do a clean install.

    What is going on?

    The system will continue startup into r5 and in fact the sound which
    appears to fail during startup is actually working. In fact, by touching /root/tmp/asound.state.tmp I think maybe I got rid of that alsa ctl error message, but haldaemon still takes forever to start up and still reports itself as failed on bootup. Something else possibly related to that--there are over one hundred udev processes running shortly after booting up! The udev entries in the process table dwindle down to just one after a while, but hald continues to hog 95% or more of the cpu. The system is usable after booting: X starts up (including aiglx +
    beryl or compiz, etc.) But this booting situation is screwed up and totally unacceptable. I wasn't sure if system cpu idle was somehow mixed into hal number because it is so high. I can find no messages in the system log or the bootup log that explain or even relate to this hald craziness. Also when you add things like USB devices to the running system, it takes much longer for them to be recognized than before with MDV 2006.0

    System details include amd XP 2500+ Barton core, Nforce2 motherboard, 512 MB DDR, 2 ide harddrives, an ide cdrom, and a Yamaha scsi cdrw (plus
    Advansys ultrascsi host adaptor)

    Thanks in Advance for any help

    --
    Get Big Brother out of my email to reply

  2. Re: Problem with Hal on 2007.1

    On Mon, 07 May 2007 15:20:12 +0000, hazzmat wrote:

    > Hal is taking 97-99% of cpu and stalls the bootup process for several minutes!
    >
    >


    Ok I tried stopping haldaemon (it doesn't obey the service stop command,
    requiring a killall -9, (very nice!)) and then I tried running it manually
    with --daemon=no --verbose=yes and logging stderr to a file. I let that
    run for just a couple of minutes, until I saw that the log file hald was
    creating was already over 80 megabytes in size. Over 99% of that file are
    lines like this
    13:05:18.065 [I] device.c:3504: add_dev: subsys=scsi_host sysfs_path=/sys/device
    s/platform/host0 dev= parent_dev=0x00000000

    exactly like that differeing only by timestamp.

    Yes, it seems hald is borken w respect to scsi hostadapters. What a POS.
    Any ideas about how to salvage this situation?
    thanks.
    --
    Get Big Brother out of my email to reply

  3. Re: Problem with Hal on 2007.1

    On Mon, 07 May 2007 17:10:19 +0000, hazzmat wrote:

    I will have to check whether the scsi cdrw is terminated. It's never not
    worked before so...
    Won't be able to get to that til tomorrow, meanwhile does anyone have a
    different idea I should check out?

    --
    Get Big Brother out of my email to reply

  4. Re: Problem with Hal on 2007.1

    On Mon, 07 May 2007 13:10:19 -0400, hazzmat wrote:

    > Yes, it seems hald is borken w respect to scsi hostadapters. What a POS.
    > Any ideas about how to salvage this situation?


    Until you do figure out what the problem is, I'd stop hal from running
    at start up. "chkconfig haldaemon off" (as root).

    Sorry I can't be of more help, I haven't encountered this problem before.

    Regards, Dave Hodgins

    --
    Change nomail.afraid.org to ody.ca to reply by email.
    (nomail.afraid.org has been set up specifically for
    use in usenet. Feel free to use it yourself.)

  5. Re: Problem with Hal on 2007.1

    On Mon, 07 May 2007 17:10:19 +0000, hazzmat wrote:

    I checked the termination jumper of the scsi cdrw today, and it is correct
    (it's never been moved in years). So I then bought a termination adaptor
    for the external bus of the card. So the scsi bus is absolutely and
    definitely terminated, internally and externally, but that didn't help.
    Removing the scsi adaptor from the system removes the problem from the
    haldaemon startup. Hal will then no longer stall and fail and doesn't take
    all the cpu with the scsi stuff removed. But then there's no scsi or cdrw.

    Turns out this bug (or one closely related) has been seen by mandriva
    users and reported to qa.mandriva.com. Two weeks ago, Frederic Crozat of
    Mandriva said their kernel hacker had reproduced it. It's known to adoring
    fans as Bug #30310.
    http://qa.mandriva.com/show_bug.cgi?id=30310

    If anyone else has seen this bug please go to the above url for bug 30310
    at qa.mandriva.com and vote for it to be fixed.

    Thank You.


    --
    Get Big Brother out of my email to reply

  6. Re: Problem with Hal on 2007.1

    On Mon, 07 May 2007 17:10:19 +0000, hazzmat wrote:

    > On Mon, 07 May 2007 15:20:12 +0000, hazzmat wrote:
    >
    >> Hal is taking 97-99% of cpu and stalls the bootup process for several
    >> minutes!
    >>
    >>
    >>


    They now have some replacement rpms in cooker that solve the hang during
    bootup problem, but unless I'm missing something, it's no real fix: hald
    just IGNORES the scsi hostadapter altogether and the cdrw that is attached
    through it.

    --
    Get Big Brother out of my email to reply

+ Reply to Thread