[O/S] dev-loop devices causing long boot time?

[davdklkn]

I was trying to optimize my boot time a bit.

The commands

Code: Select all

systemd-analyze critical-chain
systemd-analyze blame

systemd-analyze appear to blame plymouth-quit-wait.service for an increased boot time:

Code: Select all

$ systemd-analyze blame

6.573s plymouth-quit-wait.service
1.636s snapd.service
1.362s e2scrub_reap.service
1.098s fwupd.service
 959ms udisks2.service
 928ms dev-sdc4.device
 823ms containerd.service
 655ms systemd-journal-flush.service
 644ms ModemManager.service
 556ms snapd.seeded.service
 555ms systemd-udevd.service
 447ms accounts-daemon.service
 425ms gdm.service
 381ms polkit.service
 ...

Code: Select all

$ systemd-analyze critical-chain

...
graphical.target @10.936s
└─multi-user.target @10.936s
  └─plymouth-quit-wait.service @4.361s +6.573s
    └─systemd-user-sessions.service @4.321s +28ms
      └─network.target @4.288s
        └─wpa_supplicant.service @4.063s +223ms
          └─dbus.service @3.791s +242ms
            └─basic.target @3.780s
              └─sockets.target @3.780s
                └─snapd.socket @3.757s +22ms
                  └─sysinit.target @3.753s
                    └─snapd.apparmor.service @3.589s +164ms
                      └─apparmor.service @3.280s +290ms
                        └─local-fs.target @3.279s
                          └─run-credentials-systemd\x2dtmpfiles\x2dsetup.service.mount @3.289s
                            └─local-fs-pre.target @1.306s
                              └─systemd-tmpfiles-setup-dev.service @1.281s +24ms
                                └─systemd-sysusers.service @1.252s +28ms
                                  └─systemd-remount-fs.service @1.212s +26ms
                                    └─systemd-journald.socket @1.180s
                                      └─system.slice @1.159s
                                        └─-.slice @1.159s

As far as I understand however, plymouth-quit-wait.service is just responsible for showing a boot animation and does that only until all other boot tasks are done.

So then I did

Code: Select all

systemd-analyze plot > boot_analysis.svg

and if I understand correctly, plymouth-quit-wait.service at the bottom

does indeed wait for other tasks further up:

So apparently some virtual devices like dev-loop0.device cause a big chunk of my startup time..

After some googling I learned that they might have something to do with snapd packages?

I have also made other plots, where the same devices are initialized much faster:

Am I understanding correctly, that the dev-loop devices contribute a lot to my startup time? (And that plymouth-quit-wait.service shouldn't have anything to do with it?)
Why could it be, that they appear to start much faster in the plot sometimes? Am I overlooking/misunderstanding something?
How can I prevent those devices taking so much time without removing snapd?


Any help is much appreciated!
Thanks!