Switched to systemd, now 2 minutes added to boot time [Solved]

[paul1149]

Charlie,

Just compare the uuids with fstab:

Actually, when I set up fstab, a year or so ago, I discovered the very neat ability to identify drives by labels rather than UUIDs. Very user friendly. Of course, the labels should be unique, but that's very easy. So my fstab looks like this:

Code: Select all

LABEL=Swap-500	none	swap	discard	0	0

This method comes with one caveat, though - one must use the correct label in fstab! And letter case matters!

... But an attempt at hibernation failed. ...rebooted rather than restoring the previous session...

Code: Select all

lsblk -f | grep swap

Code: Select all

echo RESUME=UUID=XXX | sudo tee /etc/initramfs-tools/conf.d/resume ; sudo update-initramfs -uk all

Reboot.

I did come across something to that effect. Then I discovered that I don't have the resume file, so I wondered. So today I created `resume` and ran the command:

Code: Select all

root@mx:/etc/initramfs-tools/conf.d# echo RESUME=LABEL=Swap-500 | sudo tee /etc/initramfs-tools/conf.d/resume ; sudo update-initramfs -uk all
RESUME=LABEL=Swap-500
update-initramfs: Generating /boot/initrd.img-6.1.0-13-amd64
W: Possible missing firmware /lib/firmware/amdgpu/ip_discovery.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/vega10_cap.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/sienna_cichlid_cap.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/navi12_cap.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/psp_13_0_10_ta.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/psp_13_0_10_sos.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/aldebaran_cap.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_imu.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_rlc.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_mec.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_me.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_pfp.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_0_toc.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/sdma_6_0_3.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/sienna_cichlid_mes1.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/sienna_cichlid_mes.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/navi10_mes.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_4_mes_2.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_mes1.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_mes_2.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_3_mes.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_2_mes_2.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_1_mes_2.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/gc_11_0_0_mes_2.bin for module amdgpu
W: Possible missing firmware /lib/firmware/amdgpu/smu_13_0_10.bin for module amdgpu
root@mx:/etc/initramfs-tools/conf.d# 

As you can see, lots of potential problems, but hibernate is working. I left a text file with some unsaved text in it, and when it came back up the unsaved text was preserved.

So victory! Thanks much!

[Charlie Brown]

Ok, but you can still use the uuid (no matter label is used in fstab and not uuid)... Yes, label has such disadvantages :)

In case you decide to do it that way, just redoing the command with uuid is enough, the new file will overwrite the previous one.

(I would change both fstab and resume, takes just 1-2 minutes) .. up to you :)

[paul1149]

No, no, no. I'm very happy with the label approach. It's a blessing.

[Charlie Brown]

In the meantime, those "Possible missing firmware .." things are ignorable (when everything's working).. Or you can either download that part (amdgpu) or the whole firmware from github and put in place ( /lib/firmware/... ) :)

Manually, just amdgpu:

https://git.kernel.org/pub/scm/linux/ke ... ree/amdgpu

or the whole (the most recent)

• Ctrl+Alt+T

• Code: Select all

  wget https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/snapshot/linux-firmware-20231211.tar.gz ; tar -xf linux-firmware-20231211.tar.gz ; cd linux-firmware-20231211 ; sudo cp -R * /lib/firmware/ ; sudo update-initramfs -uk all

Reboot

If you like only :)

[paul1149]

Yeah, I should keep a copy of it, but as long as it's working I'm going to leave it be. Thanks.

[Charlie Brown]

Ok, you're welcome :)

[asqwerth]

This may happen again whenever you install a distro on your machine, and you chose to format the existing swap partition afresh instead of simply reusing swap or choosing no swap during installation ( and then editing fstab later).

If you format swap partition afresh, the uuid changes, but fstab file still records old swap uuid. In systemd, this leads to the 90 sec disk check during boot up.

[paul1149]

I was surprised that systemd picked up the fstab error, but sysV did not. Not sure why that is. But in any case it's resolved, and now I have a swap system that actually functions.

[asqwerth]

I was surprised that systemd picked up the fstab error, but sysV did not. Not sure why that is. But in any case it's resolved, and now I have a swap system that actually functions.

It's a "feature" of systemd to have the 90sec check if fstab and actual uuids don't match.

When Sysvinit finds mismatches, it simply does not mount fstab-listed partitions whose uuids can't be found.

This has been the case since systemd became a thing (I have many other systemd distros in different partitions). So I make sure I reuse existing swap upon installation, not format it again.

[paul1149]

Ah, ok, thanks. I guess it's good in that it alerted me to a hidden problem. But it seems to me that a simple visual alert to an fstab situation would have been better than trying to square root what was causing a delay.