[SOLVED] Updated F16 hangs during start of X?

JZL240I-U · 03-27-2012, 08:26 AM

After the initial installation of F16 everything ran smoothly from the POST messages to login. Then the fedora team published some updates and since then progress stops when the central "f" is nearly finished. I then have to hit repeatedly <Ctrl><Alt><Backspace> to come to the login screen. What might cause that and how to find a remedy?

macemoneta · 03-27-2012, 02:46 PM

Hit escape when the Fedora logo first appears to see the text console progress and any errors that may be occurring.

JZL240I-U · 03-29-2012, 04:28 AM

Did that and it runs through

. So I restarted without hitting <Esc> and waited until the "roadblock" shows up. Anter a long wait I hit <Esc>, the screen changes and the boot-messages whisk along till the presumtive road block appears. That reads

Code:

Starting Sendmail Mail Transport Client

After some more waiting comes

Code:

Started Sendmail Mail Transport Client
Failed to start LSB
{some more (about 5 lines) very fast messages}

And then the login screen.

Okay, I could simply always hit <Esc> or edit the GRUB commands (will have to look that up, I'm new to GRUB2), but what could cause sendmail to block the boot sequence? And why can <Ctrl><Alt><Backspace> break the lock?

macemoneta · 03-29-2012, 10:33 AM

Try running as root:

Code:

systemd-analyze blame | less

It will tell you what's taking so long in the boot. Hitting ctrl-alt-backspace before X starts does nothing but waste time until the boot continues.

JZL240I-U · 03-29-2012, 10:42 AM

Quote:

Originally Posted by macemoneta

...Hitting ctrl-alt-backspace before X starts does nothing but waste time until the boot continues.

I'm quite sure it is not. First time I left the room for some time to do something else and looked back in to see whether the login screen had appeared. No chance. So far only ctrl-alt-backspace re-animates the boot sequence -- or escape and the flow of the messages.

I tried ctrl-alt-backspace on a hunch: the little diagonally growing central pear with the Fedora "f" is all black and white and shows no "f" on a blue background as it did earlier. That`s why I thought it is a graphics problem.

Be that as it may, I'll try your suggestions and report later.

<edit> Wrong. It's not X at all. See post #8. </edit>

JZL240I-U · 03-30-2012, 11:41 AM

Okay, this is the output of "systemd-analyze blame | less" of the services needing more than 0.5 seconds to start:

Code:

  8022ms udev-settle.service
  2469ms fedora-loadmodules.service
  1658ms media.mount
  1604ms remount-rootfs.service
  1589ms dev-mqueue.mount
  1572ms dev-hugepages.mount
  1560ms sys-kernel-security.mount
  1553ms sys-kernel-debug.mount
  1422ms mdmonitor-takeover.service
  1401ms udev-trigger.service
  1353ms udev.service
  1267ms systemd-remount-api-vfs.service
  1267ms systemd-vconsole-setup.service
  1141ms sm-client.service
  1093ms sys-kernel-config.mount
  1004ms fedora-storage-init-late.service
   871ms fedora-readonly.service
   799ms NetworkManager.service
   687ms systemd-tmpfiles-setup.service
   632ms avahi-daemon.service
   598ms rsyslog.service
   581ms fedora-wait-storage.service
...
   102ms sendmail.service
...

I have no idea what to look for, does this look normal? At least sendmail doesn't seem to be the culprit.

Oh, btw. I updated the system yesterday (~480 MB of updates). The system starts now after about 90 seconds. Darn sluggish, but if this is the way it is...

<edit> Found the bastard: I issued a "systemd-analyze plot > bootchart.svg" command and looked at the resulting plot and lo and behold "plymouth-quit-wait.service" starts from seconds <24 to >89. What can I do about that?

macemoneta · 03-30-2012, 02:08 PM

Your 'systemd-analyze blame' shows nothing delaying the start. The plymouth-quit-wait.service is normal; it's just waiting to remove the graphic. At this point, you can use:

Code:

systemctl --full --type=service list-units

This will give you a list of the started system services. If there are any that you don't use, you can disable them. For example:

Code:

systemctl disable sendmail.service

JZL240I-U · 04-02-2012, 02:15 AM

Bullseye.

After the latest update I get during boot something like

Code:

...
Starting LSB: Starts and Stops login and scanning of iSCSI devices
Starting Sendmail Mail Transport Client
Started Sendmail Mail Transport Client
Failed to start LSB
...
{about five more lines}
{...and the login screen}

I tried to verify that with dmesg -- no luck. So I used both commands you provided and found out that there are two services connected to this, namely "iscsi.service" and "iscsid.service". I disabled both and erverything is back to normal.

To sum it up: You were right on all accounts and you provided all the information I needed to solve this. Thanks and kudos macemoneta, you were right on the mark up to the point of my erroneously suspecting X for all the hassle

.