• niceload slackpkg upgrade-all

    From Tom Crane@Use-Author-Supplied-Address-Header@[127.1] to alt.os.linux.slackware on Wed Oct 14 20:03:14 2026
    From Newsgroup: alt.os.linux.slackware

    Can anyone comment on whether this works on your Slackware system?

    I am running Slackware64-current and have no problem doing 'slackpkg upgrade-all' but want to control the system load due to running
    updates.

    The process starts OK, does the initial checks but then pauses
    indefinitely. Tracing it with 'niceload sh -x /usr/sbin/slackpkg
    upgrade-all' shows the problem is in the dialog application which
    displays the list of packages to upgrade and allows the user to select
    packages to exclude -- a feature I use. The last few lines of the
    shell trace before the pause are,

    + cat /tmp/slackpkg.qAUTay/dialog.tmp
    + xargs dialog --title upgrade --backtitle 'slackpkg 16.0.1' --item-help --checklist 'Choose packages to upgrade:' 19 70 13

    The problem isn't the system load rising to the point where niceload
    kicks in and stops the processes running under it. Using 'niceload -L
    <large number>' or 'niceload -v' demonstrates that.

    Thanks
    Tom

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From kaukasoina3dore73js4@kaukasoina3dore73js4@sci.fi (Petri Kaukasoina) to alt.os.linux.slackware on Sun Oct 18 09:40:57 2026
    From Newsgroup: alt.os.linux.slackware

    Tom Crane <Use-Author-Supplied-Address-Header@[127.1]> wrote:
    Can anyone comment on whether this works on your Slackware system?

    It doesn't.

    Have you tried this:

    chrt -i ionice -c 3 slackpkg upgrade-all
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Tom Crane@Use-Author-Supplied-Address-Header@[127.1] to alt.os.linux.slackware on Thu Oct 22 15:45:43 2026
    From Newsgroup: alt.os.linux.slackware

    Petri Kaukasoina <kaukasoina3dore73js4@sci.fi> wrote:
    Tom Crane <Use-Author-Supplied-Address-Header@[127.1]> wrote:
    Can anyone comment on whether this works on your Slackware system?

    It doesn't.

    Have you tried this:

    chrt -i ionice -c 3 slackpkg upgrade-all

    Thanks. Glad to know its not just my system.

    I was not aware of chrt, so thanks. 'chrt -i ionice -c 3' is useful in
    keeping the system load down but I still got the loadavg rising to
    above 8 during the update process with 8+ processes and kernel threads
    in the d-wait state at times.

    I had a quick browse of the GNU Parallel documentation (of which
    niceload is a part) but could not find any bug reports or caveats
    relating to running dialog.

    Regards
    Tom Crane

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Henrik Carlqvist@Henrik.Carlqvist@deadspam.com to alt.os.linux.slackware on Fri Oct 23 05:49:56 2026
    From Newsgroup: alt.os.linux.slackware

    On Thu, 22 Oct 2026 15:45:43 +0000, Tom Crane wrote:
    I still got the loadavg rising to above 8 during the update process
    with 8+ processes and kernel threads in the d-wait state at times.

    I have not used slackpkg myself, but basically, Slackware packages are
    really simple compressed tar files. How can you get a load higher than 8
    by installing or upgrading packages?

    Usually packages tools like installpkg and upgradepkg have checks to make
    sure that they are not being run while another package is being installed
    or upgraded. Those checks are implemented with calls to flock and some of
    them can be overrided by a package by having the magic word "NOLOCK" in
    the file install/doinst.sh.

    Newer versions of xz do support parallell decompression by default, but
    that would give you processes in running state (R) rather than waiting
    for disk (D).

    Is it really so that those more than 8 processes waiting for Disk are processes from your package installation/upgrade? Or could it be that you
    are running somethhing else (like a parallell compilation of some big
    sourc tree) and that "something else" is what is causing your high load?

    All those tricks with nice, ionice and chrt will not be able to lower an existing load. In the best case, those tricks will only avoid adding
    even more to an existing high load.

    If you really are getting those high loads waiting for disk by installing
    or upgrading Slackware packages my only explanation is that you are
    installing or upgrading multiple packages in parallell and that seems
    like a rather bad idea. The original Slackware package management tools
    like installpkg and upgradepkg have mechanisms to prevent parallell installations or upgrades.

    regards Henrik
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From kaukasoina3dore73js4@kaukasoina3dore73js4@sci.fi (Petri Kaukasoina) to alt.os.linux.slackware on Fri Oct 23 12:23:20 2026
    From Newsgroup: alt.os.linux.slackware

    Tom Crane <Use-Author-Supplied-Address-Header@[127.1]> wrote:
    I was not aware of chrt, so thanks. 'chrt -i ionice -c 3' is useful in >keeping the system load down but I still got the loadavg rising to
    above 8 during the update process with 8+ processes and kernel threads
    in the d-wait state at times.

    How many cores do you have? (If eight then isn't it good to use them?)
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Marco Moock@mm@dorfdsl.de to alt.os.linux.slackware on Fri Oct 23 16:05:21 2026
    From Newsgroup: alt.os.linux.slackware

    Am 23.10.26 um 07:49 schrieb Henrik Carlqvist:
    On Thu, 22 Oct 2026 15:45:43 +0000, Tom Crane wrote:
    I still got the loadavg rising to above 8 during the update process
    with 8+ processes and kernel threads in the d-wait state at times.

    I have not used slackpkg myself, but basically, Slackware packages are
    really simple compressed tar files. How can you get a load higher than 8
    by installing or upgrading packages?

    On very slow systems, this is possible, as some bigger packets (kernel,
    X11 stuff, initrd) need some resources. Checksums are also checked etc.

    The TO should specify the hardware.
    --
    Gru|f
    Marco

    Spam bitte an abfalleimer2001@stinkedores.dorfdsl.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Rich@rich@example.invalid to alt.os.linux.slackware on Fri Oct 23 19:05:30 2026
    From Newsgroup: alt.os.linux.slackware

    Petri Kaukasoina <kaukasoina3dore73js4@sci.fi> wrote:
    Tom Crane <Use-Author-Supplied-Address-Header@[127.1]> wrote:
    I was not aware of chrt, so thanks. 'chrt -i ionice -c 3' is useful in >>keeping the system load down but I still got the loadavg rising to
    above 8 during the update process with 8+ processes and kernel threads
    in the d-wait state at times.

    How many cores do you have? (If eight then isn't it good to use them?)

    Not if all eight are fighting over the same mechanical disk [1] holding
    the root filesystem. The random seeks resulting from eight packages
    being installed at the same time would slow the process to a crawl.


    [1] Yes, this assumes OP has a mechanical disk. OP has never said
    which type root disk is in use.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Henrik Carlqvist@Henrik.Carlqvist@deadspam.com to alt.os.linux.slackware on Sat Oct 24 10:06:29 2026
    From Newsgroup: alt.os.linux.slackware

    On Fri, 23 Oct 2026 16:05:21 +0200, Marco Moock wrote:
    On very slow systems, this is possible, as some bigger packets (kernel,
    X11 stuff, initrd) need some resources. Checksums are also checked etc.

    Yes, some big packages need more CPU cycles to get decompressed and of
    course they also need to spend more time reading from the installation/
    upgrade directory and writing to the system disk. Performing checksum
    also of course takes its time.

    But more than 8 different processes waiting for Disk? At what phase of a package installation can you get to that? The package installation
    scripts will need 1 process to read the package and 1 process to write to
    the system disk. I can only come to think of two different scenarios when
    a package installation would leave you with more than 8 processes waiting
    for disk:

    1) The installation script /install/doinst.sh is doing some parallell
    disk IO. Writing such an installation script for a Slackware package
    seems like a rather bad idea to me.

    2) The system has used so much RAM that it has started swapping. If so,
    not only the package installation processes will be waiting for disk, but
    also other processes. That would mean that most of the 8+ processes
    waiting for disk are not related to the package installation/upgrade.

    regards Henrik
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Rich@rich@example.invalid to alt.os.linux.slackware on Sat Oct 24 21:07:18 2026
    From Newsgroup: alt.os.linux.slackware

    Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
    On Fri, 23 Oct 2026 16:05:21 +0200, Marco Moock wrote:
    On very slow systems, this is possible, as some bigger packets (kernel,
    X11 stuff, initrd) need some resources. Checksums are also checked etc.

    Yes, some big packages need more CPU cycles to get decompressed and of course they also need to spend more time reading from the installation/ upgrade directory and writing to the system disk. Performing checksum
    also of course takes its time.

    But more than 8 different processes waiting for Disk? At what phase of a package installation can you get to that? The package installation
    scripts will need 1 process to read the package and 1 process to write to the system disk.

    The OP, possibly in a later post in this thread, mentioned GNU
    Parallel. This implies they may be attempting to install multiple
    packages in parallel via parallel or xargs in "parallel jobs mode".
    This is the only way I can see, beyond the op opening eight xterms and manually starting eight install loops for different package groups.


    --- Synchronet 3.22a-Linux NewsLink 1.2