How to restart a process after zypper patch

golson765 · November 16, 2012, 5:14am

Howdy all,

I just did a zypper patch, and then I ran zypper ps to see what needed to be restarted. I was wanting to figure out how to set everything right so that my system is not running on any deleted processes, but I want to do it without a reboot.

So I logged out of KDE and logged back in (should fix all the kde processes, right?), and then I ran zypper patch again. Here are the results:

george@tribaltrekker:~> zypper ps
The following running processes use deleted files:


 

 george@tribaltrekker:~> zypper ps
 The following running processes use deleted files:
 

 PID  | PPID | UID  | Login  | Command        | Service | Files                      
 -----+------+------+--------+----------------+---------+----------------------------
 3940 | 1    | 1000 | george | alsa-source    |         | /usr/lib64/libuuid.so.1.3.0
 6677 | 1    | 1000 | george | winedevice.exe |         | /usr/lib/libuuid.so.1.3.0  
 6709 | 1    | 1000 | george | wine-preloader |         | /usr/lib/libuuid.so.1.3.0  
 6711 | 1    | 1000 | george | wine-preloader |         | /usr/lib/libuuid.so.1.3.0

So now I am thinking, how do I restart these processes with a command line command? I know how to end the processes with a kill signal, but not how to restart one, or even what is requiring these processes. When I search for the process on the system activity gui, it gives me the parent process, so I suppose I can use the system activity gui to restart those.

But I was wondering how this could be done with a command line. If I have to restart the entire parent process, it would be good to know how to use the command line to figure out what is the correct parent process also.

Ultimately the goal is to be able to do this over ssh, and I know that doing graphical over ssh makes things really slow, especially from where I am. So I want to make sure I know how to do it with the command line.

Also, apparently a process is different from a system service (is that right?), so the systemctl restart command in systemd won’t help. At least, that is how I understand it, please correct me if I am wrong.

Thanks in advance,

malcolmlewis · November 16, 2012, 5:40am

On Fri 16 Nov 2012 04:16:01 AM CST, golson765 wrote:

Howdy all,

I just did a zypper patch, and then I ran zypper ps to see what needed
to be restarted. I was wanting to figure out how to set everything right
so that my system is not running on any deleted processes, but I want to
do it without a reboot.

So I logged out of KDE and logged back in (should fix all the kde
processes, right?), and then I ran zypper patch again. Here are the
results:

george@tribaltrekker:~> zypper ps
The following running processes use deleted files:

Code:

george@tribaltrekker:~> zypper ps
The following running processes use deleted files:

PID | PPID | UID | Login | Command | Service |
Files
-----±-----±-----±-------±---------------±--------±---------------------------
3940 | 1 | 1000 | george | alsa-source |
| /usr/lib64/libuuid.so.1.3.0 6677 | 1 | 1000 | george |
winedevice.exe | | /usr/lib/libuuid.so.1.3.0 6709 | 1 | 1000
| george | wine-preloader | | /usr/lib/libuuid.so.1.3.0 6711 |
1 | 1000 | george | wine-preloader |

/usr/lib/libuuid.so.1.3.0

So now I am thinking, how do I restart these processes with a command
line command? I know how to end the processes with a kill signal, but
not how to restart one, or even what is requiring these processes. When
I search for the process on the system activity gui, it gives me the
parent process, so I suppose I can use the system activity gui to
restart those.

But I was wondering how this could be done with a command line. If I
have to restart the entire parent process, it would be good to know how
to use the command line to figure out what is the correct parent process
also.

Ultimately the goal is to be able to do this over ssh, and I know that
doing graphical over ssh makes things really slow, especially from
where I am. So I want to make sure I know how to do it with the command
line.

Also, apparently a process is different from a system service (is that
right?), so the systemctl restart command in systemd won’t help. At
least, that is how I understand it, please correct me if I am wrong.

Thanks in advance,

Hi
In my case I had some udev processes hang onto the same lib, a restart
of the udev systemd service removed them.

So you have some sort of wine server running? A restart of that should
help. The pstree command on the pid may help identify the starting
service or daemon;


pstree -p

Look at the output to identify the main process pid.

–
Cheers Malcolm °¿° (Linux Counter #276890)
openSUSE 12.2 (x86_64) Kernel 3.4.11-2.16-desktop
up 4 days 2:27, 4 users, load average: 0.03, 0.06, 0.05
CPU Intel i5 CPU M520@2.40GHz | Intel Arrandale GPU

golson765 · November 16, 2012, 8:45am

malcolmlewis wrote:
>
> [/QUOTE]
> Hi
> In my case I had some udev processes hang onto the same lib, a restart
> of the udev systemd service removed them.
>
> So you have some sort of wine server running? A restart of that should
> help. The pstree command on the pid may help identify the starting
> service or daemon;
>


> pstree -p
>

Look at the output to identify the main process pid.

Thanks, that is helpful. Yes, I am running some sort of wine server, so that I can use my
Evernote client.

Now this was very interesting, because how do you know the actual name of the service? For
example, when I ran zypper ps as root, I got the following:



tribaltrekker:/home/george # zypper ps
The following running processes use deleted files:

PID   | PPID | UID  | Login  | Command        | Service | Files
------+------+------+--------+----------------+---------+-----------------------------
390   | 1    | 0    | root   | udevd          |         | /usr/lib64/libblkid.so.1.1.0
|      |      |        |                |         | /usr/lib64/libuuid.so.1.3.0
801   | 1    | 0    | root   | agetty         |         | /usr/sbin/agetty (deleted)
3940  | 1    | 1000 | george | alsa-source    |         | /usr/lib64/libuuid.so.1.3.0
6677  | 1    | 1000 | george | winedevice.exe |         | /usr/lib/libuuid.so.1.3.0
6709  | 1    | 1000 | george | wine-preloader |         | /usr/lib/libuuid.so.1.3.0
6711  | 1    | 1000 | george | wine-preloader |         | /usr/lib/libuuid.so.1.3.0
8484  | 390  | 0    | root   | udevd          |         | /usr/lib64/libblkid.so.1.1.0
|      |      |        |                |         | /usr/lib64/libuuid.so.1.3.0
11582 | 390  | 0    | root   | udevd          |         | /usr/lib64/libblkid.so.1.1.0
|      |      |        |                |         | /usr/lib64/libuuid.so.1.3.0

So it turns out that udev is an issue on my machine as well. But note that the service column is
blank, and that the command for udev is “udevd”. When I ran “systemctl status udevd.service”, I
got an error saying “udevd.service” did not exist. But when I ran “systemctl status
udev.service”, it gave me information about the udev service. So the name of the service is
“udev” without the trailing “d”.

So, with udev, that is a pretty big service, so it should probably be common knowledge that the
name of the service is “udev” without any trailing “d”. However, with other services, how do you
find the actual name of the service so that you can restart it?

When I did pstree on my alsa-source pid, here was the result:



tribaltrekker:/home/george # pstree -s 3940
systemd───pulseaudio───4*{pulseaudio}]
tribaltrekker:/home/george # systemctl status pulseaudio.service
pulseaudio.service
Loaded: error (Reason: No such file or directory)
Active: inactive (dead)

So you can see that pulseaudio is the culprit, but how do I find out the actual name of the
pulseaudio service so that I can restart it using “systemctl restart <name>.service”?

malcolmlewis · November 16, 2012, 1:27pm

On Fri 16 Nov 2012 07:45:51 AM CST, golson765 wrote:

malcolmlewis wrote:
>
>

> Hi
> In my case I had some udev processes hang onto the same lib, a restart
> of the udev systemd service removed them.
>
> So you have some sort of wine server running? A restart of that should
> help. The pstree command on the pid may help identify the starting
> service or daemon;
>


> pstree -p
>

Look at the output to identify the main process pid.

Thanks, that is helpful. Yes, I am running some sort of wine server, so
that I can use my Evernote client.

Now this was very interesting, because how do you know the actual name
of the service? For example, when I ran zypper ps as root, I got the
following:



tribaltrekker:/home/george # zypper ps
The following running processes use deleted files:

PID   | PPID | UID  | Login  | Command        | Service |
Files
------+------+------+--------+----------------+---------+-----------------------------
390   | 1    | 0    | root   | udevd          |
| /usr/lib64/libblkid.so.1.1.0 |      |      |        |
|         | /usr/lib64/libuuid.so.1.3.0 801   | 1    | 0    | root   |
agetty         |         | /usr/sbin/agetty (deleted) 3940  | 1    |
1000 | george | alsa-source    |         | /usr/lib64/libuuid.so.1.3.0
6677  | 1    | 1000 | george | winedevice.exe |
| /usr/lib/libuuid.so.1.3.0 6709  | 1    | 1000 | george |
wine-preloader |         | /usr/lib/libuuid.so.1.3.0 6711  | 1    |
1000 | george | wine-preloader |         | /usr/lib/libuuid.so.1.3.0
8484  | 390  | 0    | root   | udevd          |
| /usr/lib64/libblkid.so.1.1.0 |      |      |        |
|         | /usr/lib64/libuuid.so.1.3.0 11582 | 390  | 0    | root   |
udevd          |         | /usr/lib64/libblkid.so.1.1.0 |      |
|        |                |         | /usr/lib64/libuuid.so.1.3.0

So it turns out that udev is an issue on my machine as well. But note
that the service column is blank, and that the command for udev is
“udevd”. When I ran “systemctl status udevd.service”, I got an error
saying “udevd.service” did not exist. But when I ran “systemctl status
udev.service”, it gave me information about the udev service. So the
name of the service is “udev” without the trailing “d”.

So, with udev, that is a pretty big service, so it should probably be
common knowledge that the name of the service is “udev” without any
trailing “d”. However, with other services, how do you find the actual
name of the service so that you can restart it?

When I did pstree on my alsa-source pid, here was the result:



tribaltrekker:/home/george # pstree -s 3940
systemd───pulseaudio───4*{pulseaudio}]
tribaltrekker:/home/george # systemctl status pulseaudio.service
pulseaudio.service
Loaded: error (Reason: No such file or directory)
Active: inactive (dead)

So you can see that pulseaudio is the culprit, but how do I find out
the actual name of the pulseaudio service so that I can restart it
using “systemctl restart <name>.service”?

[/QUOTE]
Hi
Just run the command systemctl list-units to see what’s loaded.

If you just run the pstree -p command to see the whole
tree then should get back further rather than isolate one pid which may
be out on the branch rather at the trunk

–
Cheers Malcolm °¿° (Linux Counter #276890)
openSUSE 12.2 (x86_64) Kernel 3.4.11-2.16-desktop
up 4 days 10:11, 4 users, load average: 0.00, 0.01, 0.05
CPU Intel i5 CPU M520@2.40GHz | Intel Arrandale GPU

robin_listas · November 16, 2012, 2:14pm

On 2012-11-16 08:45, golson765 wrote:

>
> So it turns out that udev is an issue on my machine as well. But note that the service column is
> blank, and that the command for udev is “udevd”. When I ran “systemctl status udevd.service”, I

At this point, I do not care: I do an “init 1, init 5” sequence.

–
Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 “Celadon” (Minas Tirith))

golson765 · November 16, 2012, 2:18pm

malcolmlewis wrote:

>

> Hi
> Just run the command systemctl list-units to see what’s loaded.
>
> If you just run the pstree -p command to see the whole
> tree then should get back further rather than isolate one pid which may
> be out on the branch rather at the trunk

Hmm, interesting. For pulseaudio, the next thing up on the tree hierarchy is systemd itself. Same
for winedevice.

Still don’t know how I would restart pulseaudio from the command line. I could kill the process,
but then how do you get it going again? Apparently there is no specific “service” that governs
it? So its parent process is systemd, and you can’t restart systemd without a reboot, can you?

robin_listas · November 16, 2012, 2:38pm

On 2012-11-16 14:18, golson765 wrote:

> Hmm, interesting. For pulseaudio, the next thing up on the tree hierarchy is systemd itself. Same
> for winedevice.

For pulse I log out from all graphical sessions. Then I log in text mode
as user, check if it is running, and kill it. It is restarted as I log
in graphical mode again.

–
Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 “Celadon” (Minas Tirith))

malcolmlewis · November 16, 2012, 2:38pm

On Fri 16 Nov 2012 01:18:24 PM CST, golson765 wrote:

malcolmlewis wrote:

> [QUOTE]

> Hi
> Just run the command systemctl list-units to see what’s loaded.
>
> If you just run the pstree -p command to see the whole
> tree then should get back further rather than isolate one pid which
> may be out on the branch rather at the trunk

Hmm, interesting. For pulseaudio, the next thing up on the tree
hierarchy is systemd itself. Same for winedevice.

Still don’t know how I would restart pulseaudio from the command line.
I could kill the process, but then how do you get it going again?
Apparently there is no specific “service” that governs it? So its
parent process is systemd, and you can’t restart systemd without a
reboot, can you?

[/QUOTE]
Hi
You could try sending a signal to the process (probably check
pulseaudio and wine documentation), to get a list of signals use the
command kill -l. Else as Carlos pointed out, drop to runlevel 1 and
back to 5…

–
Cheers Malcolm °¿° (Linux Counter #276890)
openSUSE 12.2 (x86_64) Kernel 3.4.11-2.16-desktop
up 4 days 11:26, 4 users, load average: 0.01, 0.03, 0.05
CPU Intel i5 CPU M520@2.40GHz | Intel Arrandale GPU

arvidjaar · November 16, 2012, 4:40pm

If you are using systemd, you can use systemd-cgls to get overview of all services and processes that belong to services. To find out service for one particular process you can use ps:

bor@opensuse:~/src/grub> pgrep -l udevd
331 udevd
28402 udevd
28456 udevd
bor@opensuse:~/src/grub> ps -o cgroup -p 331,28402,28456
CGROUP
3:cpuacct,cpu:/system/udev.service,1:name=systemd:/system/udev.service
3:cpuacct,cpu:/system/udev.service,1:name=systemd:/system/udev.service
3:cpuacct,cpu:/system/udev.service,1:name=systemd:/system/udev.service
bor@opensuse:~/src/grub> pgrep -l pulseaudio
11795 pulseaudio
bor@opensuse:~/src/grub> ps -o cgroup -p 11795
CGROUP
1:name=systemd:/user/bor/47
bor@opensuse:~/src/grub>

In this case pulseaudio is not a service, but is started as part of user session as is indicated by /user hierarchy as opposed to /system. And you can look at all processes in this session using systemd-loginctl:

bor@opensuse:~/src/grub> systemd-loginctl --no-pager session-status 47
47 - bor (501)
       Since: Tue, 13 Nov 2012 19:07:31 +0400; 3 days ago
      Leader: 11689 (gdm-session-wor)
        Seat: seat0; vc8
     Display: :0
     Service: gdm; type x11; class user
      Active: yes
      CGroup: /user/bor/47
          ├ 11689 gdm-session-worker [pam/gdm]
          ├ 11701 /usr/bin/gnome-keyring-daemon --daemonize --login
          ├ 11720 /usr/bin/gnome-session
          ├ 11765 dbus-launch --exit-with-session /usr/bin/gnome-ses...
          ├ 11766 /bin/dbus-daemon --fork --print-pid 5 --print-addr...
          ├ 11768 /usr/lib/gvfs/gvfsd
          ├ 11770 /usr/lib/gvfs//gvfs-fuse-daemon -f /run/user/bor/g...
          ├ 11784 /usr/lib/gnome-settings-daemon-3.0/gnome-settings-...
          ├ 11795 /usr/bin/pulseaudio --start --log-target=syslog
          ├ 11798 /usr/lib/pulse/gconf-helper
          ├ 11800 /usr/lib/GConf/2/gconfd-2
          ├ 11802 /usr/lib/gnome-settings-daemon-3.0/gsd-printer
          ├ 11807 /usr/lib/gvfs/gvfs-udisks2-volume-monitor
          ├ 11810 /usr/lib/gvfs/gvfs-afc-volume-monitor
          ├ 11813 /usr/lib/gvfs/gvfs-gphoto2-volume-monitor
          ├ 11821 /usr/bin/gnome-shell
          ├ 11823 syndaemon -i 1.0 -K -R -t
          ├ 11877 /usr/lib/tracker-store
          ├ 11879 /usr/bin/python /usr/share/system-config-printer/a...
          ├ 11891 gnome-screensaver
          ├ 11896 /usr/lib/tracker-miner-fs
          ├ 11904 /usr/lib/evolution/3.4/evolution-alarm-notify
          ├ 11906 nm-applet
          ├ 11914 /usr/lib/gnome-shell/gnome-shell-calendar-server
          ├ 11916 /usr/lib/mission-control-5
          ├ 11919 /usr/lib/evolution-data-server/evolution-calendar-...
          ├ 11926 /usr/lib/goa-daemon
          ├ 11951 /usr/lib/evolution-data-server/evolution-addressbo...
          ├ 11961 /usr/lib/gvfs/gvfsd-afc --spawner :1.1 /org/gtk/gv...
          ├ 11963 /usr/lib/gvfs/gvfsd-afc --spawner :1.1 /org/gtk/gv...
          ├ 11969 /usr/lib/libsocialweb-core
          ├ 11991 gnome-terminal
          ├ 11997 gnome-pty-helper
          ├ 11998 bash
          ├ 12045 claws-mail
          ├ 12059 /usr/lib64/firefox/firefox
          ├ 12076 /usr/lib/at-spi2/at-spi-bus-launcher
          ├ 12101 /usr/lib/gvfs/gvfsd-http --spawner :1.1 /org/gtk/g...
          ├ 12229 bash
          ├ 14983 /usr/lib/dconf-service
          ├ 20677 bash
          └ 28911 systemd-loginctl --no-pager session-status 47