LinuxQuestions.org - Mad error (services) go LVM

- SUSE / openSUSE (https://www.linuxquestions.org/questions/suse-opensuse-60/)

- - Mad error (services) go LVM (https://www.linuxquestions.org/questions/suse-opensuse-60/mad-error-services-go-lvm-282144/)

Mad error: Stopped services activate LVM

Switching off services activates LVM in SuSE 9.0pro?

This is impossible.

I tried to streamline the boot sequence of my stand-alone workstation (no net-(card), just modem) and thus set the services nscd, portmap, sshd to "off" for all runlevels and service xdm runlevels 3 and 5 to "on". I used chkconfig, no fancy tools.

After two or three boots with varying messages I got the following error during boot (note: I don't use LVM, just a "stable" system with ext2):

Code:

.

.

.

Kernel logging (ksyslog) stopped.

Kernel log daemon terminating.



Boot logging started on /dev/tty1(/dev/console) at Tue Jan 25 12:33:03 2005



<notice>run boot scripts (setserial boot.proc)

Configuring serial ports...



[several lines deleted]



<notice>run boot scripts (boot.md)

<notice>exit status of (boot.md) is (0)

<notice>run boot scripts (boot.lvm)

Run file system check on root for LVM activation

fsck 1.34 (25-Jul-2003)

/dev/hdb5 is mounted.



^G^G^G^GWARNING!!!  Running e2fsck on a mounted filesystem may cause

SEVERE filesystem damage.^G^G^G



Do you really want to continue (y/n)? no



check aborted.

doneRemounting root file system (/) read/write for vgscan...

Scanning for LVM volume groups...

vgscan -- reading all physical volumes (this may take a while...)

vgscan -- "/etc/lvmtab" and "/etc/lvmtab.d" successfully created

vgscan -- WARNING: This program does not do a VGDA backup of your volume group



Activating LVM volume groups...

vgchange -- no volume groups found



done

<notice>exit status of (boot.lvm) is (0)

<notice>run boot scripts (boot.localfs)

Activating swap-devices in /etc/fstab...

doneChecking file systems...

fsck 1.34 (25-Jul-2003)

/dev/hdb5 has been mounted 20 times without being checked, check forced.

.

.

.

I'm aware that there shouldn't be any causal or even logical connection to this from just using chkconfig. (BTW /dev/hdb5 is /).

In desperation I manually reset everything to the original values, to no avail. Then I restored the backup from chkconfig which I had done imediately before starting this. Again, the error stayed with me (including changing error messages during the first three boots, now stable).

I tried to find the script which is calling boot.lvm -- no luck.
I tried to find the script which is calling vgscan and producing this message -- no luck.
I tried to find a file in /etc with a new time stamp -- no luck.
I tried YAST, but the LVM-manager knows of no volume group.

How can I get rid of this error?
This is interrupting every boot sequence by requesting extra manual input...*argh*

This is impossible... :mad:

*bump*

@mods: should this be in a different forum (i.e. maybe this is not SuSE-specific)? In case you get active, could you please change the stupid name I gave the thread in my first exitement to the caption in the first line? Thanks a lot...

Okay, for those who are interested, I found a solution and some (but not all) answers.

In brief: don't use chkconfig with SuSE, use YAST's runlevel editor instead. There are some runlevels in SuSE which chkconfig doesn't display. If you do use chkconfig, run SuSEConfig after it -- I didn't do this, what probably caused the erratic behaviour concerning the boot messages ant the beginning. Oh, and SuSE says you have to run SuSEConfig...

Long explanation: After the excitement died down, I found what goes on on boot time :rolleyes:. There is also a SuSE README in /etc/init.d/README on this topic.

After the Kernel has done its thing it hands control to "init". /sbin/init controls the next steps of the boot process, first by evaluating /etc/inittab which tells it what runlevel to use.

Most important, though, it calls /etc/init.d/boot.d in which the runlevel is finally evaluated (the number is used to determine the runlevel-specific directory which hosts links to the start scripts which are started in a loop from here). And before that is a loop which controls runlevel "b", like boot, (SuSE-specific?) and kicks off boot.lvm, my nemesis.

Btw. /etc/init.d/boot.d also reads /etc/rc.status (which sets the pretty colors for SuSE's boot-screen) and /etc/sysconfig.boot (where the behaviour of the boot process can be customized).

Okay, so I used YAST to power down boot.lvm and got rid of the error.

BUT -- I never found out what activated boot.lvm in the first place, or, if it was active all the time, what caused the error.

Note for those who want to accelerate their boot sequence: chkconfig can delete or insert links to services for boot-levels 1 through 5. It does not deinstall the module / service -- but YAST can...