Very slow boot, "waiting for /dev/mapper/fedora_xxx"

I’m seeing upwards of 90 seconds delay on boot. systemd-analyze ambiguously blames systemd-udev-settle, lvm2-monitor, and lvm2-pvscan@8:4 (until I masked lvm2-monitor, which barely improved performance but it now blames only systemd-udev-settle.) Setting use_lvmetad = 0 in lvm.conf seemed to help slightly, too, but there seems to be an unavoidable 30-second wait.

Here is lvm2-pvscan@8:4 from journalctl:

-- Reboot --
May 18 11:03:53 localhost.localdomain systemd[1]: Starting LVM event activation on device 8:4...
May 18 11:03:53 localhost.localdomain lvm[732]:   pvscan[732] PV /dev/sda4 online, VG fedora_localhost is complete.
May 18 11:03:53 localhost.localdomain lvm[732]:   pvscan[732] VG fedora_localhost run autoactivation.
May 18 11:04:24 localhost.localdomain lvm[732]:   3 logical volume(s) in volume group "fedora_localhost" now active
May 18 11:04:24 localhost.localdomain systemd[1]: Finished LVM event activation on device 8:4.
-- Reboot --
May 19 04:03:31 localhost.localdomain systemd[1]: Starting LVM event activation on device 8:4...
May 19 04:03:31 localhost.localdomain lvm[731]:   pvscan[731] PV /dev/sda4 online, VG fedora_localhost is complete.
May 19 04:03:31 localhost.localdomain lvm[731]:   pvscan[731] VG fedora_localhost run autoactivation.
-- Reboot --
May 19 09:01:40 localhost.localdomain systemd[1]: Stopping LVM event activation on device 8:4...
May 19 09:01:41 localhost.localdomain lvm[3089]:   pvscan[3089] PV /dev/sda4 online.
May 19 09:01:41 localhost.localdomain systemd[1]: lvm2-pvscan@8:4.service: Succeeded.
May 19 09:01:41 localhost.localdomain systemd[1]: Stopped LVM event activation on device 8:4.
-- Reboot --
May 19 11:29:37 localhost.localdomain systemd[1]: Starting LVM event activation on device 8:4...
May 19 11:29:37 localhost.localdomain lvm[720]:   pvscan[720] PV /dev/sda4 online, VG fedora_localhost is complete.
May 19 11:29:37 localhost.localdomain lvm[720]:   pvscan[720] VG fedora_localhost run autoactivation.
May 19 11:30:09 localhost.localdomain lvm[720]:   3 logical volume(s) in volume group "fedora_localhost" now active
May 19 11:30:09 localhost.localdomain systemd[1]: Finished LVM event activation on device 8:4.
May 21 12:28:51 localhost.localdomain systemd[1]: Stopping LVM event activation on device 8:4...
May 21 12:28:53 localhost.localdomain lvm[23633]:   pvscan[23633] PV /dev/sda4 online.
May 21 12:28:53 localhost.localdomain systemd[1]: lvm2-pvscan@8:4.service: Succeeded.

>spend eternity curating a great post
>new users can’t post more than 2 links
I will be returning from the void with advice momentarily after I figure out how to grind this bizarre forum software

facts:

  • This is a real problem [1][2]
  • It’s already been fixed [3]
  • (…just, not in RHEL or Fedora [4])
  • In fact, Red Hat, Inc. doesn’t even understand it [5][6]

tl;dr: use this backported fix for now, iff you’re affected by this issue:

https://copr.fedorainfracloud.org/coprs/oleastre/systemd/
(thanks, Oleastre!)

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.