You are not logged in.

#1 2016-10-21 15:46:57

jonnyt43
Member
Registered: 2016-10-21
Posts: 2

[SOLVED] Long boot time after system-wide update "A start job..."

Hi All,

This is my first time posting to the forum. Apologies if I get this wrong...

As indicated in my subject line above, I am experiencing a 'long' boot time after a system-wide update (I used the 'sudo pacman -Syu' command, and proceeded to update as usual). On boot I notice the following message: "A start job is running on sys-subs...s-enp2s0.device" with a moving " * " and a counter in seconds up to 1min 30s proceeds. Only when the counter has reached the 90 seconds (1min30s) does boot continue successfully.

After reading a few other posts about similar boot time problems I ran the following commands with their corresponding outputs:

$ systemd-analyze critical-path

graphical.target @1min 31.177s
└─multi-user.target @1min 31.176s
  └─smbd.service @1min 31.115s +60ms
    └─nmbd.service @1min 31.012s +87ms
      └─network.target @1min 30.998s
        └─NetworkManager.service @1.503s +62ms
          └─dbus.service @1.451s
            └─basic.target @1.450s
              └─sockets.target @1.449s
                └─avahi-daemon.socket @1.449s
                  └─sysinit.target @1.445s
                    └─systemd-update-utmp.service @1.442s +2ms
                      └─systemd-tmpfiles-setup.service @1.431s +9ms
                        └─local-fs.target @1.430s
                          └─srv-nfs4-storage.mount @1.429s +1ms
                            └─storage.mount @1.345s +64ms
                              └─systemd-fsck@dev-disk-by\x2duuid-22e0aafe\x2d2f2
                                └─dev-disk-by\x2duuid-22e0aafe\x2d2f22\x2d4e73\x

$ systemd-analyze blame

289ms dev-sda1.device
           176ms systemd-fsck@dev-disk-by\x2duuid-22e0aafe\x2d2f22\x2d4e73\x2db8
           174ms systemd-journald.service
            87ms nmbd.service
            77ms polkit.service
            65ms systemd-journal-flush.service
            64ms storage.mount
            63ms upower.service
            62ms NetworkManager.service
            61ms systemd-fsck@dev-disk-by\x2duuid-505520f9\x2d5527\x2d4a03\x2d84
            60ms smbd.service
            54ms alsa-restore.service
            50ms systemd-udevd.service
            47ms udisks2.service
            38ms avahi-daemon.service
            36ms systemd-modules-load.service
            27ms accounts-daemon.service
            25ms systemd-tmpfiles-setup-dev.service
            24ms systemd-udev-trigger.service
            22ms kmod-static-nodes.service
            22ms systemd-remount-fs.service
            22ms sys-kernel-debug.mount
            22ms lm_sensors.service
            20ms dev-mqueue.mount
            17ms home.mount
            17ms user@1000.service
            14ms systemd-binfmt.service
            13ms systemd-tmpfiles-clean.service
            12ms colord.service
            11ms dev-hugepages.mount
            10ms dev-sda2.swap
             9ms systemd-tmpfiles-setup.service
             7ms systemd-sysctl.service
             6ms systemd-user-sessions.service
             6ms systemd-vconsole-setup.service
             6ms proc-sys-fs-binfmt_misc.mount
             5ms systemd-random-seed.service
             4ms systemd-logind.service
             2ms rtkit-daemon.service
             2ms systemd-update-utmp.service
             2ms nvidia-persistenced.service
             2ms sys-fs-fuse-connections.mount
             1ms tmp.mount
             1ms sys-kernel-config.mount
             1ms srv-nfs4-storage.mount

$ systemctl --failed
0 loaded units listed.

$ systemd-analyze
Startup finished in 1.444s (kernel) + 1min 31.177s (userspace) = 1min 32.622s

Can anyone help me resolve this long boot time please? Any help will be gratefully received!!

Best regards

Jon

Last edited by jonnyt43 (2016-10-21 16:18:04)

Offline

#2 2016-10-21 15:59:06

loqs
Member
Registered: 2014-03-06
Posts: 17,362

Re: [SOLVED] Long boot time after system-wide update "A start job..."

Offline

#3 2016-10-21 16:15:31

jonnyt43
Member
Registered: 2016-10-21
Posts: 2

Re: [SOLVED] Long boot time after system-wide update "A start job..."

Dear loqs,

It worked!!

Thank you so much!

For anyone else with this issue, loqs's link indicated I needed to do the following:

$ cd /usr/lib/systemd/system

then

$ sudo nano systemd-user-sessions.service

which opens nano and has the following:

     #  This file is part of systemd.
     #
     #  systemd is free software; you can redistribute it and/or modify it
     #  under the terms of the GNU Lesser General Public License as published by
     #  the Free Software Foundation; either version 2.1 of the License, or
     #  (at your option) any later version.

     [Unit]
     Description=Permit User Sessions
     Documentation=man:systemd-user-sessions.service(8)
     After=remote-fs.target nss-user-lookup.target network.target

     [Service]
     Type=oneshot
     RemainAfterExit=yes
     ExecStart=/usr/lib/systemd/systemd-user-sessions start
     ExecStop=/usr/lib/systemd/systemd-user-sessions stop
     ######

I then proceeded to delete "network.target" in the line that reads:  "After= ..... network.target"

Made sure I wrote-out the changes and exited. I then rebooted, success, no delay!

SOLVED!!

Last edited by jonnyt43 (2016-10-21 16:19:36)

Offline

#4 2016-10-23 18:35:38

vyu223
Member
Registered: 2013-02-18
Posts: 24

Re: [SOLVED] Long boot time after system-wide update "A start job..."

Just checking - is there a reason that you're editing the unit directly in /usr/lib/systemd/system?

Did you instead mean to do something like editing /etc/systemd/system/systemd-user-sessions.service.d/override.conf and setting as its contents:

[Unit]
After=
After=remote-fs.target nss-user-lookup.target

Edit: For reference (from the systemd arch wiki page): "To avoid conflicts with pacman, unit files provided by packages should not be directly edited."

Last edited by vyu223 (2016-10-23 18:40:14)

Offline

#5 2016-10-23 19:03:54

loqs
Member
Registered: 2014-03-06
Posts: 17,362

Re: [SOLVED] Long boot time after system-wide update "A start job..."

@vyu223 does that work for you?

man 5 systemd.unit wrote:

       Note that dependencies (After=, etc.) cannot be reset to an empty list,
       so dependencies can only be added in drop-ins. If you want to remove
       dependencies, you have to override the entire unit.

Which caused me to have to override the entire unit.
I agree about the path my post p1632295 after modification is meant to show that /usr/lib/systemd/system/systemd-user-sessions.service is incorrect hence the strike through and use /etc/systemd/system/systemd-user-sessions.service instead.
I did not want to simply replace the wrong path with the correct one and be accused of hiding my errors.

Offline

#6 2016-10-23 22:39:01

vyu223
Member
Registered: 2013-02-18
Posts: 24

Re: [SOLVED] Long boot time after system-wide update "A start job..."

It does work for me! On my up-to-date system, that man page says the same thing for me as it does for you. Maybe it needs to be updated? I haven't looked into the commits for systemd to see if there was a change that started allowing this.

I do remember, though, that in one point in time, we were not allowed to clear the ExecStart array by doing "ExecStart=". And then some time later, we started being able to do it. Maybe something similar has happened for "After=".

I was originally directing my post toward the OP; I did notice in your post that you changed the path to /etc/systemd/system. Sorry for not being more clear.

EDIT: Huh. It must have been a fluke. I tried restarting today and had the original problem the OP had again. It looks like the man page is still accurate; the After array cannot be cleared by doing "After=". Overrode the whole unit (instead of using the drop in file) and have my system booting up fine now.

Last edited by vyu223 (2016-10-24 16:24:25)

Offline

Board footer

Powered by FluxBB