[opensuse] Why do we reboot after a systemd upgrade instead of --system daemon-reexec?
General question, On 15.2 today, updates included systemd, ghostscript and a few other peripheral libraries, but nothing other than systemd would have conceivably called for a reboot, but yet zypper up said a reboot was required. Question, why when we update systemd do we call for a reboot instead of executing a --system daemon-reexec in the post-install part of the update? E.g. systemctl --system daemon-reexec That itself will reload system and any/all new service files. More a curiosity than a problem. -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
* David C. Rankin
General question,
On 15.2 today, updates included systemd, ghostscript and a few other peripheral libraries, but nothing other than systemd would have conceivably called for a reboot, but yet zypper up said a reboot was required.
Question, why when we update systemd do we call for a reboot instead of executing a --system daemon-reexec in the post-install part of the update? E.g.
systemctl --system daemon-reexec
That itself will reload system and any/all new service files. More a curiosity than a problem.
I don't believe systemd changes require a reboot, but dbus and kernel changes do, and perhaps other I do not know. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Wed, Aug 5, 2020 at 3:28 AM David C. Rankin
General question,
On 15.2 today, updates included systemd, ghostscript and a few other peripheral libraries, but nothing other than systemd would have conceivably called for a reboot, but yet zypper up said a reboot was required.
Question, why when we update systemd do we call for a reboot instead of executing a --system daemon-reexec in the post-install part of the update? E.g.
systemctl --system daemon-reexec
Using the example of automounted file systems defined in /etc/fstab that are managed by systemd, will this command cause all the various support scripts to also be rerun? I never seem to get changes to work as expected without a reboot. But I have not used daemon-reexec, only daemon-reload -- Roger Oberholtzer -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 8/5/20 2:17 AM, Roger Oberholtzer wrote:
Using the example of automounted file systems defined in /etc/fstab that are managed by systemd, will this command cause all the various support scripts to also be rerun?
I never seem to get changes to work as expected without a reboot. But I have not used daemon-reexec, only daemon-reload
I have never not had an fstab entry not continue to work following a daemon-reexec. Arch went to using daemon-reexec as part of it's systemd package updates almost two years ago. The daemon-reload only covers making changes to .service files available to systemd, the daemon-reexec is just that a re exec of systemd so it will invoke everything that systemd did to begin with (absent whatever magic the systemd folks coded to make it go right) Having lived with it for a couple of years on Arch (and no problems at all with all server processes, multiple raid arrays, etc..) I was wondering if we could do something similar to avoid the "Reboot needed" message -- when a reboot wasn't really required? -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
* David C. Rankin
On 8/5/20 2:17 AM, Roger Oberholtzer wrote:
Using the example of automounted file systems defined in /etc/fstab that are managed by systemd, will this command cause all the various support scripts to also be rerun?
I never seem to get changes to work as expected without a reboot. But I have not used daemon-reexec, only daemon-reload
I have never not had an fstab entry not continue to work following a daemon-reexec.
Arch went to using daemon-reexec as part of it's systemd package updates almost two years ago. The daemon-reload only covers making changes to .service files available to systemd, the daemon-reexec is just that a re exec of systemd so it will invoke everything that systemd did to begin with (absent whatever magic the systemd folks coded to make it go right)
Having lived with it for a couple of years on Arch (and no problems at all with all server processes, multiple raid arrays, etc..) I was wondering if we could do something similar to avoid the "Reboot needed" message -- when a reboot wasn't really required?
please provide an example of an instance where "reboot needed" was provided and the reboot was not needed. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 8/5/20 6:07 AM, Patrick Shanahan wrote:
please provide an example of an instance where "reboot needed" was provided and the reboot was not needed.
I'll get the zypper log or list of packages updated last night on 15.2 and post those later today. That will give us a baseline. -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 8/5/20 2:28 PM, David C. Rankin wrote:
On 8/5/20 6:07 AM, Patrick Shanahan wrote:
please provide an example of an instance where "reboot needed" was provided and the reboot was not needed.
I'll get the zypper log or list of packages updated last night on 15.2 and post those later today. That will give us a baseline.
Here are the packages that were updated on 8/4 and none that I can tell would require a reboot, but nonetheless zypper stated Reboot Required: # sed -n 's#^2020-08-04.*/\(\w\w*.*[.]rpm\)$#\1#p' < /var/log/zypper.log | sort -u diffutils-3.6-lp152.5.3.1.x86_64.rpm diffutils-lang-3.6-lp152.5.3.1.noarch.rpm ghostscript-9.52-lp152.2.4.1.x86_64.rpm ghostscript-x11-9.52-lp152.2.4.1.x86_64.rpm htop-2.2.0-lp152.3.6.x86_64.rpm libraw16-0.18.9-lp152.5.3.1.x86_64.rpm libsystemd0-234-lp152.31.4.1.x86_64.rpm libsystemd0-32bit-234-lp152.31.4.1.x86_64.rpm libudev1-234-lp152.31.4.1.x86_64.rpm systemd-234-lp152.31.4.1.x86_64.rpm systemd-bash-completion-234-lp152.31.4.1.noarch.rpm systemd-network-234-lp152.31.4.1.x86_64.rpm systemd-sysvinit-234-lp152.31.4.1.x86_64.rpm udev-234-lp152.31.4.1.x86_64.rpm All can be updated without a reboot, the only two of consequence, systemd and udev can be handled with: systemctl --system daemon-reexec udevadm control --reload-rules && udevadm trigger So can we do something like this to minimize the number of reboots zypper calls for? Except for kernel updates, everything else should be able to be done without a reboot (there may be some others I'm not aware of), but systemd and udev shouldn't be two. -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
David C. Rankin wrote:
So can we do something like this to minimize the number of reboots zypper calls for?
I suspect it is best discussed on the factory list. I am certainly no expert, but I think the need-reboot flag is determined by /etc/zypp/needreboot which lists packages that require reboot (includes systemd and udev). -- Per Jessen, Zürich (17.6°C) http://www.dns24.ch/ - your free DNS host, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 8/6/20 2:04 AM, Per Jessen wrote:
David C. Rankin wrote:
So can we do something like this to minimize the number of reboots zypper calls for?
I suspect it is best discussed on the factory list.
I am certainly no expert, but I think the need-reboot flag is determined by /etc/zypp/needreboot which lists packages that require reboot (includes systemd and udev).
Yes, that is why I was so surprised when after updating the listed packages I was faced with a reboot prompt. Nothing needed a reboot as systemd and udev could have been handled with two simple commands instead of a reboot. -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
David C. Rankin wrote:
On 8/6/20 2:04 AM, Per Jessen wrote:
David C. Rankin wrote:
So can we do something like this to minimize the number of reboots zypper calls for?
I suspect it is best discussed on the factory list.
I am certainly no expert, but I think the need-reboot flag is determined by /etc/zypp/needreboot which lists packages that require reboot (includes systemd and udev).
Yes, that is why I was so surprised when after updating the listed packages I was faced with a reboot prompt.
Were you actually asked "reboot yes/no ?" ? I just seem to get the suggestion that the changes require a reboot.
Nothing needed a reboot as systemd and udev could have been handled with two simple commands instead of a reboot.
Yesterday I patched a system running mariadb. The mariadb patch included an upgrade script which unfortunately failed and caused mariadb not to start. Automagically restarting things isn't always a good idea. (just playing devils advocate). -- Per Jessen, Zürich (19.2°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 8/6/20 3:15 AM, Per Jessen wrote:
Were you actually asked "reboot yes/no ?" ? I just seem to get the suggestion that the changes require a reboot.
No, I was just told "Rebbot needed" -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 8/6/20 9:15 AM, David C. Rankin wrote:
On 8/5/20 2:28 PM, David C. Rankin wrote:
On 8/5/20 6:07 AM, Patrick Shanahan wrote:
please provide an example of an instance where "reboot needed" was provided and the reboot was not needed.
I'll get the zypper log or list of packages updated last night on 15.2 and post those later today. That will give us a baseline.
Here are the packages that were updated on 8/4 and none that I can tell would require a reboot, but nonetheless zypper stated Reboot Required:
# sed -n 's#^2020-08-04.*/\(\w\w*.*[.]rpm\)$#\1#p' < /var/log/zypper.log | sort -u diffutils-3.6-lp152.5.3.1.x86_64.rpm diffutils-lang-3.6-lp152.5.3.1.noarch.rpm ghostscript-9.52-lp152.2.4.1.x86_64.rpm ghostscript-x11-9.52-lp152.2.4.1.x86_64.rpm htop-2.2.0-lp152.3.6.x86_64.rpm libraw16-0.18.9-lp152.5.3.1.x86_64.rpm libsystemd0-234-lp152.31.4.1.x86_64.rpm libsystemd0-32bit-234-lp152.31.4.1.x86_64.rpm libudev1-234-lp152.31.4.1.x86_64.rpm systemd-234-lp152.31.4.1.x86_64.rpm systemd-bash-completion-234-lp152.31.4.1.noarch.rpm systemd-network-234-lp152.31.4.1.x86_64.rpm systemd-sysvinit-234-lp152.31.4.1.x86_64.rpm udev-234-lp152.31.4.1.x86_64.rpm
All can be updated without a reboot, the only two of consequence, systemd and udev can be handled with:
systemctl --system daemon-reexec udevadm control --reload-rules && udevadm trigger
So can we do something like this to minimize the number of reboots zypper calls for? Except for kernel updates, everything else should be able to be done without a reboot (there may be some others I'm not aware of), but systemd and udev shouldn't be two.
dbus is certainly one the current implementation has no way to track connections across restarts. -- Simon Lees (Simotek) http://simotek.net Emergency Update Team keybase.io/simotek SUSE Linux Adelaide Australia, UTC+10:30 GPG Fingerprint: 5B87 DB9D 88DC F606 E489 CEC5 0922 C246 02F0 014B
participants (5)
-
David C. Rankin
-
Patrick Shanahan
-
Per Jessen
-
Roger Oberholtzer
-
Simon Lees