About an issue of run-postinsts.service

About an issue of run-postinsts.service

[Maohui Lei (Fujitsu)]

Hi, 

After systemd has been upgraded to version 255.*, if users add a package with a postsints script(e.g. gnutls-fips), the following warning would output when execute systemd in the first boot.
-----------------------------------------------------------------------
root@qemux86-64:~# systemctl status syslog.socket
* syslog.socket - Syslog Socket
     Loaded: loaded (/usr/lib/systemd/system/syslog.socket; static)
     Active: active (running) since Wed 2024-05-22 06:06:41 UTC; 5min ago
   Triggers: * busybox-syslog.service
       Docs: man:systemd.special(7)
             https://www.freedesktop.org/wiki/Software/systemd/syslog
     Listen: /run/systemd/journal/syslog (Datagram)
     CGroup: /system.slice/syslog.socket

Notice: journal has been rotated since unit was started, output may be incomplete.
Warning: The unit file, source configuration file or drop-ins of syslog.socket changed on disk. Run 'systemctl daemon-reload' to reload units.
-----------------------------------------------------------------------

I found that the option "--no-reload" in run-postinsts.service causes the issue. Systemd255 needs to reload unites even "enable/disable" units.
------------------------------------------------------------------------
root@qemux86-64:~# systemctl --version
systemd 255 (255.4^)

root@qemux86-64:~# systemctl --no-reload disable sshd.socket
Removed "/etc/systemd/system/sockets.target.wants/sshd.socket".
root@qemux86-64:~# systemctl status sshd.socket

sshd.socket
   Loaded: loaded (/usr/lib/systemd/system/sshd.socket; disabled; preset: enabled)
   Active: active (listening) since Thu 2024-05-16 02:11:17 UTC; 8min ago
   Listen: [::]:22 (Stream)
 Accepted: 0; Connected: 0;
  Process: 193 ExecStartPre=/usr/bin/mkdir -p /var/run/sshd (code=exited, status=0/SUCCESS)
    Tasks: 0 (limit: 258)
   Memory: 584.0K (peak: 1.4M)
      CPU: 389ms
   CGroup: /system.slice/sshd.socket

May 16 02:11:17 qemux86-64 systemd[1]: Starting sshd.socket...
May 16 02:11:17 qemux86-64 systemd[1]: Listening on sshd.socket.
Warning: The unit file, source configuration file or drop-ins of sshd.socket changed on disk. Run 'systemctl daemon-reload' to reload units.
------------------------------------------------------------------------

And I tested by qemu and on PC, there is no issue that mentioned in the patch(fd266b1b66844d660ebd8e33fe22c08c8ddaccbf) which added "--no-reload " in run-postinsts.service.
So, can I submit a patch to delete the "--no-reload " from run-postinsts.service?

Best regards
Lei

Re: [OE-core] About an issue of run-postinsts.service

[Alexander Kanavin]

On Thu, 23 May 2024 at 05:00, leimaohui via lists.openembedded.org
<leimaohui=fujitsu.com@lists.openembedded.org> wrote:

> And I tested by qemu and on PC, there is no issue that mentioned in the patch(fd266b1b66844d660ebd8e33fe22c08c8ddaccbf) which added "--no-reload " in run-postinsts.service.
> So, can I submit a patch to delete the "--no-reload " from run-postinsts.service?

I think yes you should. The original change feels like a hack around
bugs in systemd, those bugs may have been solved long time ago, and if
they have not, we need a better way to handle this (that does not
result in systemd warnings).

Alex

Re: [OE-core] About an issue of run-postinsts.service

[Max]

Hi

Am Donnerstag, dem 23.05.2024 um 08:23 +0200 schrieb Alexander Kanavin via lists.openembedded.org:
> On Thu, 23 May 2024 at 05:00, leimaohui via lists.openembedded.org
> <leimaohui=fujitsu.com@lists.openembedded.org> wrote:
> 
> > And I tested by qemu and on PC, there is no issue that mentioned in the patch(fd266b1b66844d660ebd8e33fe22c08c8ddaccbf) which added "--no-reload " in run-postinsts.service.
> > So, can I submit a patch to delete the "--no-reload " from run-postinsts.service?
> 
> I think yes you should. The original change feels like a hack around
> bugs in systemd, those bugs may have been solved long time ago, and if
> they have not, we need a better way to handle this (that does not
> result in systemd warnings).
> 

We since removed any pkg_postinstall_ontarget from our images built with OE master.

I now build an image with a dummy postinstall added to see if the behavior can still be reproduced.
That simple test on an i.MX6 based SoM no longer shows the issue, so likely it was fixed in systemd
/ OE / wherever.

I'm fine with the change to remove "--no-reload " in the run-postinsts.service.

Thanks for pinging us Alex.

Regards
Max

> 
> Alex
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#199778): https://lists.openembedded.org/g/openembedded-core/message/199778
> Mute This Topic: https://lists.openembedded.org/mt/106256417/3617484
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [max.oss.09@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
> 

RE: [OE-core] About an issue of run-postinsts.service

[Maohui Lei (Fujitsu)]

Hi, 

Thank you for your reply. So, I'll submit a patch to remove "--no-reload" from run-postinsts.service.


Best regards
Lei

> -----Original Message-----
> From: Max <max.oss.09@gmail.com>
> Sent: Friday, May 24, 2024 1:14 AM
> To: alex.kanavin@gmail.com; Lei, Maohui/雷 茂慧 <leimaohui@fujitsu.com>
> Cc: openembedded-core@lists.openembedded.org; Max Krummenacher
> <max.krummenacher@toradex.com>; stefan.agner@toradex.com
> Subject: Re: [OE-core] About an issue of run-postinsts.service
> 
> Hi
> 
> Am Donnerstag, dem 23.05.2024 um 08:23 +0200 schrieb Alexander Kanavin
> via lists.openembedded.org:
> > On Thu, 23 May 2024 at 05:00, leimaohui via lists.openembedded.org
> > <leimaohui=fujitsu.com@lists.openembedded.org> wrote:
> >
> > > And I tested by qemu and on PC, there is no issue that mentioned in the
> patch(fd266b1b66844d660ebd8e33fe22c08c8ddaccbf) which added
> "--no-reload " in run-postinsts.service.
> > > So, can I submit a patch to delete the "--no-reload " from
> run-postinsts.service?
> >
> > I think yes you should. The original change feels like a hack around
> > bugs in systemd, those bugs may have been solved long time ago, and if
> > they have not, we need a better way to handle this (that does not
> > result in systemd warnings).
> >
> 
> We since removed any pkg_postinstall_ontarget from our images built with
> OE master.
> 
> I now build an image with a dummy postinstall added to see if the behavior
> can still be reproduced.
> That simple test on an i.MX6 based SoM no longer shows the issue, so likely
> it was fixed in systemd / OE / wherever.
> 
> I'm fine with the change to remove "--no-reload " in the run-postinsts.service.
> 
> Thanks for pinging us Alex.
> 
> Regards
> Max
> 
> >
> > Alex
> > -=-=-=-=-=-=-=-=-=-=-=-
> > Links: You receive all messages sent to this group.
> > View/Reply Online (#199778):
> > https://lists.openembedded.org/g/openembedded-core/message/199778
> > Mute This Topic: https://lists.openembedded.org/mt/106256417/3617484
> > Group Owner: openembedded-core+owner@lists.openembedded.org
> > Unsubscribe:
> https://lists.openembedded.org/g/openembedded-core/unsub
> > [max.oss.09@gmail.com]
> > -=-=-=-=-=-=-=-=-=-=-=-
> >