[SOLVED!]Very, very slow boot

[Erik]

Yesterday I decided to install Mageia 8 on a Sony Vaio VGN-FW51JF. Some time ago Ihad Mg8 on iy, but then tried Linux Mint. OK, back to Mageia 8. I started using Linux in 2004 with Mandrake 10 until about a year ago when I tried Linux Mint. I put Mageia-8-Live-Plasma-x86_64.iso on an USB stick and to my surprise starting te installer took several minutes. Has something changed since the introduction of Mg8? Have a look at 'systemd-analyze blame':

26.036s drakx-installer-xsetup.service
17.464s mandriva-everytime.service
11.146s network-up.service
3.197s shorewall.service
3.195s systemd-udev-settle.service
2.599s lvm2-monitor.service
1.916s dev-loop0.device
1.686s shorewall6.service
1.655s udisks2.service
1.299s upower.service
1.255s network.service
925ms resolvconf.service
789ms sensord.service
735ms avahi-daemon.service
733ms bluetooth.service

I never saw anything like this during all the years I used Mageia or one of its predecessors.

After installation the boot process also took several minutes (5 minutes!):

# systemd-analyze blame
17.917s home.mount
16.727s udisks2.service
14.090s network-up.service
12.821s network.service
11.563s polkit.service
5.034s lvm2-monitor.service
4.456s systemd-udev-settle.service
4.135s upower.service
3.832s dev-sdb2.device
3.542s avahi-daemon.service
3.540s bluetooth.service
3.516s dbus.service
3.485s mga-bg-res.service
3.440s rtkit-daemon.service
3.414s systemd-logind.service
3.385s shorewall.service
3.352s partmon.service
2.089s sensord.service
1.898s preload.service
1.852s plymouth-start.service
1.735s resolvconf.service
1.620s mandriva-everytime.service
1.134s cups.service
936ms systemd-tmpfiles-setup-d

I also tried installing on another laptop; the problem is the same.

The old installation, an update from Mageia 7 gives this result:

# systemd-analyze blame
10.011s network-up.service
1.971s postfix.service
1.073s udisks2.service
869ms shorewall.service
863ms network.service
716ms dev-sdb1.device
703ms mandriva-everytime.service
450ms upower.service

I have no idea what is wrong.

Erik

[Germ]

viewtopic.php?f=8&t=13515

That solved it for me.

link fixed
thanks morgano

[morgano]

Germ, your link is missing a couple digits

[doktor5000]

systemd-analyze blame only shows half of the truth. Would be useful if you could run

Code: Select all

systemd-analyze plot > boot.svg

and later attach boot.svg here.

[Erik]

Dear Doktor5000,

thank you for your help. the result of the command you taught me, is this:

https://drive.google.com/file/d/1xwWRee ... sp=sharing

I am curious to what you conclude from this.

Erik

[sturmvogel]

Changed my comment as your first upload was access protected. Works now.

[doktor5000]

I don't see several minute boot, after ~30 seconds your desktop starts, which is not very slow.

Apart from that, a few seconds can probably be shaved off here and there, but you're missing quite a few bits of context information about your system.
How does your system differ from a default installation? Did you install on an SSD or a regular HDD, what filesystem did you chose and any special stuff, like LVM or RAID or such?
Also, what's your network connection?

[sturmvogel]

Hm, the data you provided shows that this machine is using Kernel 5.10.16-desktop-1.mga8 with a system date from Feb 13 2021. That would mean that there was never an update applied and the system time is completely wrong.

Is this an up to date system or is this the output from a system started from live DVD?

[isadora]

Dear friends,

thank you so much for your help!!

I read everything you suggested. First of all I connected the clock to the internet. In the Link that Germ suggested, I found the command "journalctl -ab" and tried it. In all journals of the times I booted I found the following:

Code: Select all

mei 27 19:09:07 localhost.localdomain kernel: clocksource: timekeeping watchdog on CPU0: Marking clocksource 'tsc-early' as unstable because the skew is too large:
mei 27 19:09:07 localhost.localdomain kernel: clocksource:                       'hpet' wd_now: 87904494 wd_last: e084a2 mask: ffffffff
mei 27 19:09:07 localhost.localdomain kernel: clocksource:                       'tsc-early' cs_now: d0c191bf0 cs_last: cc75988a0 mask: ffffffffffffffff
mei 27 19:09:07 localhost.localdomain kernel: tsc: Marking TSC unstable due to clocksource watchdog
mei 27 19:09:07 localhost.localdomain kernel: TSC found unstable after boot, most likely due to broken BIOS. Use 'tsc=unstable'.
mei 27 19:09:07 localhost.localdomain kernel: sched_clock: Marking unstable (1691648508, 235492)<-(1711420465, -19536495)
mei 27 19:09:07 localhost.localdomain kernel: clocksource: Switched to clocksource hpet
..................................
mei 27 19:09:47 localhost.localdomain chronyd[861]: System clock wrong by 157.240454 seconds
mei 27 19:12:24 localhost.localdomain chronyd[861]: System clock was stepped by 157.240454 seconds

The number of seconds varies every time, but is always somewhere around two to three minutes.

the advice in the log file is: Use 'tsc=unstable'. But I do not have the faintest idea where to put this setting. Coud it be in grub.cfg? This is the only place I can think of.

I am very happy with this result, because it tells me that it is a hardware matter, not a Mageia problem. I advised my brother in Italy to install Mageia on his computer, and in order to be able to help him better I did the installation myself, to refresh my memory of the installation procedure.

Thank you all very much for your prompt help and suggestions.

Greetings

Erik

[Erik]

Dear friends,

thinking about my own hunch I put "tsc=unstable" in grub.conf. It works!!!!

Greetings

Erik

[Germ]

Glad you got it sorted.