Bug#852014: desktop-base: Boot hangs if plymouth is not installed
Package: gdm Hi Marko, Le 21 janvier 2017 16:54:35 GMT+01:00, "Marko Mäkelä" a écrit : >Hi Aurélien, > >Thank you for your response, and I am sorry for filing the bug against >the wrong package. I hope that the log output in this message will help > >to narrow down the problem. But I fear that it may remain a mystery, >and >I accept that Debian unstable can sometimes be true to its name. :) > >TL;DR version: I would suggest that the systemd-based bootup should >give >up after 3 failed attempts of starting up gdm, like I believe it used >to >be in the past. Now it seems to be doing it forever. With this additional info I'll reaffect the bug to GDM for now. Gnome maintainers, feel free to further qualify this bug if it should be against yet another package… >>I'm not sure it's raised against the correct package tough, or I will >>need more information to analyze this problem. > >Unfortunately, I cannot reproduce the problem any more (even after >removing plymouth and downgrading all systemd-related packages to >232-11 >or 232-9). But see the /var/log/apt/history.log excerpt near the end of >this message. …or maybe close it if it's not reproducible or due to a known/transient error in sid fixed in between. >>>After some analysis (booting into rescue mode and entering the root >>>password, and following the instructions to view the systemd log), >>>I figured out that the fatal error was that >>>exec /bin/plymouth failed, because the program was not installed. >> >>Could you share more complete logs about that ? > >It seems that journalctl only keeps the systemd log since the system >startup. I did not attempt to record any logs when the system did not >boot beyond the single-user mode. I do have some kernel and user-space >messages in /var/log/messages from the failed startup attempts. Maybe >the real error was this one: > >Jan 20 14:39:17 hp org.gnome.Shell.desktop[1071]: /usr/bin/gnome-shell: > >error while loading shared libraries: libmutter-cogl.so: cannot open >shared object file: No such file or directory >Jan 20 14:39:17 hp gnome-session[1063]: gnome-session-binary[1063]: >WARNING: App 'org.gnome.Shell.desktop' exited with code 127 >Jan 20 14:39:17 hp gdm3: GdmDisplay: display lasted 0,106016 seconds >Jan 20 14:39:17 hp gdm3: Child process -1059 was already dead. >Jan 20 14:39:17 hp gdm3: Child process 1043 was already dead. >Jan 20 14:39:17 hp gdm3: Unable to kill session worker process >Jan 20 14:39:17 hp /usr/lib/gdm3/gdm-x-session[1090]: Unable to run X >server >… >Jan 20 14:39:57 hp gdm3: GdmDisplay: display lasted 0,083347 seconds >Jan 20 14:39:57 hp gdm3: Child process -13450 was already dead. >Jan 20 14:39:57 hp gdm3: Child process 13437 was already dead. >Jan 20 14:39:57 hp gdm3: Unable to kill session worker process >Jan 20 14:39:57 hp /usr/lib/gdm3/gdm-x-session[13480]: Unable to run X >server >Jan 20 14:39:57 hp gdm3: Child process -13480 was already dead. >Jan 20 14:39:57 hp gdm3: Child process 13465 was already dead. >Jan 20 14:39:57 hp gdm3: Unable to kill session worker process >Jan 20 14:39:57 hp org.gnome.Shell.desktop[13510]: >/usr/bin/gnome-shell: >error while loading shared libraries: libmutter-cogl.so: cannot open >shared object file: No such file or directory >Jan 20 14:39:57 hp gnome-session[13502]: gnome-session-binary[13502]: >WARNING: App 'org.gnome.Shell.desktop' exited with code 127 >Jan 20 14:39:57 hp gdm3: GdmDisplay: display lasted 0,079726 seconds >Jan 20 14:39:57 hp gdm3: Could not start command >'/usr/lib/gdm3/gdm-session-worker': Liian monta avointa tiedostoa >Jan 20 14:39:57 hp gdm3: Child process -13498 was already dead. >Jan 20 14:39:57 hp gdm3: Child process 13486 was already dead. >Jan 20 14:39:57 hp gdm3: Unable to kill session worker process > >Back in the days before systemd, a failure to start up gdm3 or xdm or >whatever would result in a text dialog after 3 or so failed attempts, >and there would be getty listening to some virtual consoles at >/dev/tty1 >to /dev/tty6. But, when the above happened, the system was seemingly >dead. It seems to me that it went into an infinite loop and eventually >run out of file descriptors (or maybe I had pressed ctrl-alt-del which >was obeyed after 1 or 2 seconds). "Liian monta avointa tiedostoa" is >the >Finnish translation of "Too many open files". > >On some occasion, I left the system there for 5 or 10 minutes, but >there >was no progress. (And on this laptop, the status LED for mass storage >activity is pretty well hidden, so I did not even notice that there was > >constant SSD activity going on.) > >In /var/log/syslog there is a bit more detail of the above >startup/shutdown loops of the gdm service: > >Jan 20 14:39:16 hp systemd[1]: Started Session c4 of user Debian-gdm. >Jan 20 14:39:16 hp kernel: [ 35.891506] iwlwifi :02:00.0: L1 >Enabled - LTR Enabled >Jan 20 14:39:16 hp kernel: [ 35.891772] iwlwifi :02:00.0: L1 >Enabled - LTR Enabled >Jan 20 14:39:16 hp system
Bug#852014: desktop-base: Boot hangs if plymouth is not installed
Hi Aurélien, Thank you for your response, and I am sorry for filing the bug against the wrong package. I hope that the log output in this message will help to narrow down the problem. But I fear that it may remain a mystery, and I accept that Debian unstable can sometimes be true to its name. :) TL;DR version: I would suggest that the systemd-based bootup should give up after 3 failed attempts of starting up gdm, like I believe it used to be in the past. Now it seems to be doing it forever. I'm not sure it's raised against the correct package tough, or I will need more information to analyze this problem. Unfortunately, I cannot reproduce the problem any more (even after removing plymouth and downgrading all systemd-related packages to 232-11 or 232-9). But see the /var/log/apt/history.log excerpt near the end of this message. After some analysis (booting into rescue mode and entering the root password, and following the instructions to view the systemd log), I figured out that the fatal error was that exec /bin/plymouth failed, because the program was not installed. Could you share more complete logs about that ? It seems that journalctl only keeps the systemd log since the system startup. I did not attempt to record any logs when the system did not boot beyond the single-user mode. I do have some kernel and user-space messages in /var/log/messages from the failed startup attempts. Maybe the real error was this one: Jan 20 14:39:17 hp org.gnome.Shell.desktop[1071]: /usr/bin/gnome-shell: error while loading shared libraries: libmutter-cogl.so: cannot open shared object file: No such file or directory Jan 20 14:39:17 hp gnome-session[1063]: gnome-session-binary[1063]: WARNING: App 'org.gnome.Shell.desktop' exited with code 127 Jan 20 14:39:17 hp gdm3: GdmDisplay: display lasted 0,106016 seconds Jan 20 14:39:17 hp gdm3: Child process -1059 was already dead. Jan 20 14:39:17 hp gdm3: Child process 1043 was already dead. Jan 20 14:39:17 hp gdm3: Unable to kill session worker process Jan 20 14:39:17 hp /usr/lib/gdm3/gdm-x-session[1090]: Unable to run X server … Jan 20 14:39:57 hp gdm3: GdmDisplay: display lasted 0,083347 seconds Jan 20 14:39:57 hp gdm3: Child process -13450 was already dead. Jan 20 14:39:57 hp gdm3: Child process 13437 was already dead. Jan 20 14:39:57 hp gdm3: Unable to kill session worker process Jan 20 14:39:57 hp /usr/lib/gdm3/gdm-x-session[13480]: Unable to run X server Jan 20 14:39:57 hp gdm3: Child process -13480 was already dead. Jan 20 14:39:57 hp gdm3: Child process 13465 was already dead. Jan 20 14:39:57 hp gdm3: Unable to kill session worker process Jan 20 14:39:57 hp org.gnome.Shell.desktop[13510]: /usr/bin/gnome-shell: error while loading shared libraries: libmutter-cogl.so: cannot open shared object file: No such file or directory Jan 20 14:39:57 hp gnome-session[13502]: gnome-session-binary[13502]: WARNING: App 'org.gnome.Shell.desktop' exited with code 127 Jan 20 14:39:57 hp gdm3: GdmDisplay: display lasted 0,079726 seconds Jan 20 14:39:57 hp gdm3: Could not start command '/usr/lib/gdm3/gdm-session-worker': Liian monta avointa tiedostoa Jan 20 14:39:57 hp gdm3: Child process -13498 was already dead. Jan 20 14:39:57 hp gdm3: Child process 13486 was already dead. Jan 20 14:39:57 hp gdm3: Unable to kill session worker process Back in the days before systemd, a failure to start up gdm3 or xdm or whatever would result in a text dialog after 3 or so failed attempts, and there would be getty listening to some virtual consoles at /dev/tty1 to /dev/tty6. But, when the above happened, the system was seemingly dead. It seems to me that it went into an infinite loop and eventually run out of file descriptors (or maybe I had pressed ctrl-alt-del which was obeyed after 1 or 2 seconds). "Liian monta avointa tiedostoa" is the Finnish translation of "Too many open files". On some occasion, I left the system there for 5 or 10 minutes, but there was no progress. (And on this laptop, the status LED for mass storage activity is pretty well hidden, so I did not even notice that there was constant SSD activity going on.) In /var/log/syslog there is a bit more detail of the above startup/shutdown loops of the gdm service: Jan 20 14:39:16 hp systemd[1]: Started Session c4 of user Debian-gdm. Jan 20 14:39:16 hp kernel: [ 35.891506] iwlwifi :02:00.0: L1 Enabled - LTR Enabled Jan 20 14:39:16 hp kernel: [ 35.891772] iwlwifi :02:00.0: L1 Enabled - LTR Enabled Jan 20 14:39:16 hp systemd[936]: Reached target Paths. Jan 20 14:39:16 hp systemd[936]: Reached target Timers. Jan 20 14:39:16 hp systemd[936]: Listening on GnuPG cryptographic agent (access for web browsers). Jan 20 14:39:16 hp systemd[936]: Listening on GnuPG network certificate management daemon. Jan 20 14:39:16 hp systemd[936]: Listening on GnuPG cryptographic agent and passphrase cache. Jan 20 14:39:16 hp systemd[936]: Listening on GnuPG cryptographic agent (ssh-agent
Bug#852014: desktop-base: Boot hangs if plymouth is not installed
Tags: moreinfo unreproducible Hi Marko. Thank you for your bug report. I'm not sure it's raised against the correct package tough, or I will need more information to analyze this problem. Le 20 janvier 2017 19:00:37 GMT+01:00, "Marko Mäkelä" a écrit : >Package: desktop-base >Version: 9.0.1 >Severity: critical >Justification: breaks the whole system > >Dear Maintainer, > >I run Debian unstable on my desktop. Today, after "apt update&&apt >upgrade" >the system failed to boot. > >After some analysis (booting into rescue mode and entering the root >password, and following the instructions to view the systemd log), >I figured out that the fatal error was that >exec /bin/plymouth failed, because the program was not installed. Could you share more complete logs about that ? >Finally, I figured out how to manually bring up the Ethernet interface >and how to install plymouth. The next reboot worked. > >The command >grep -lr plymouth /etc >suggests that the existence of files in /usr/share/plymouth/themes >led to the assumption that plymouth is available. I am filing this >bug against desktop-base because dpkg -S suggests that the directory >is associated with desktop-base. This is really unrelated to desktop-base as far as I can tell. The path you mention is used in a "case" branch specific to *buntu and Tanglu derivatives, and not for Debian. Please have a look at the contents of 05_debian_theme. >Possibly this bug should be filed against grub-common instead, because >grub-common owns the file /etc/grub.d/05_debian_theme which references >the directory /usr/share/plymouth/themes. Yes, I'd agree with that but I'm still unsure this is the real cause of your problem. >Here are some versions of packages: > >desktop-base: 9.0.1 >grub-common: 2.02~beta3-3 >systemd: 232-12 >plymouth: 0.9.2-4 > >I got the hang with systemd 232-9 and 232-11 before installing >plymouth. I have a VM with up to date stretch and installed desktop-base 9.0.1 from sid, then uninstalled plymouth* and I could not reproduce the issue. Additional information would be welcome ! Cheers, --Aurélien
Bug#852014: desktop-base: Boot hangs if plymouth is not installed
Package: desktop-base Version: 9.0.1 Severity: critical Justification: breaks the whole system Dear Maintainer, I run Debian unstable on my desktop. Today, after "apt update&&apt upgrade" the system failed to boot. After some analysis (booting into rescue mode and entering the root password, and following the instructions to view the systemd log), I figured out that the fatal error was that exec /bin/plymouth failed, because the program was not installed. Finally, I figured out how to manually bring up the Ethernet interface and how to install plymouth. The next reboot worked. The command grep -lr plymouth /etc suggests that the existence of files in /usr/share/plymouth/themes led to the assumption that plymouth is available. I am filing this bug against desktop-base because dpkg -S suggests that the directory is associated with desktop-base. Possibly this bug should be filed against grub-common instead, because grub-common owns the file /etc/grub.d/05_debian_theme which references the directory /usr/share/plymouth/themes. Here are some versions of packages: desktop-base: 9.0.1 grub-common: 2.02~beta3-3 systemd: 232-12 plymouth: 0.9.2-4 I got the hang with systemd 232-9 and 232-11 before installing plymouth. Best regards, Marko Mäkelä -- System Information: Debian Release: 9.0 APT prefers unstable APT policy: (500, 'unstable') Architecture: amd64 (x86_64) Kernel: Linux 4.9.0-1-amd64 (SMP w/4 CPU cores) Locale: LANG=fi_FI.UTF-8, LC_CTYPE=fi_FI.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /usr/bin/dash Init: systemd (via /run/systemd/system) Versions of packages desktop-base depends on: ii dpkg 1.18.18 ii librsvg2-common 2.40.16-1 desktop-base recommends no packages. Versions of packages desktop-base suggests: pn gnome | kde-standard | xfce4 | wmaker -- no debconf information
