View previous topic :: View next topic |
Author |
Message |
Rainer06 n00b
Joined: 25 Mar 2006 Posts: 13 Location: Nuremberg
|
Posted: Sun Jan 13, 2013 11:20 pm Post subject: Problem with localmount of /home on software raid partition |
|
|
After some update of my Gentoo installation, mounting /home does not work any more
and I do not understand what is going wrong because I did not change neither hardware
not Linux configuration.
Let me describe the issue step by step as part of the boot process:
1) The kernel detects the raid array
[ 1.590647] RAID conf printout:
[ 1.590650] --- level:5 rd:3 wd:3
[ 1.590652] disk 0, o:1, dev:sda2
[ 1.590655] disk 1, o:1, dev:sdc2
[ 1.590657] disk 2, o:1, dev:sdb2
[ 1.590686] md0: detected capacity change from 0 to 990191943680
2) localmount creates some error message saying
mount: special device /dev/md0p1 does not exist
3) There is a message in dmesg saying that the two Raid partitions are available
[ 86.439951] md0: p1 p2
although this is late in the boot process.
If I try mount the /home partition manually at a later stage, everything works as expected.
This indicates a problem with the runscript, in particular with the timing or dependencies
of localmount for Raid partitions. Raid partitions seem not to be very common.
Unfortunatley, I was introducing Raid partitions since long
time ago, and it is not possible to change the system now to run without Raid partitions.
Could somebody help with this problem?
Let me provide some background info:
A)
# rc-status -a
Runlevel: nonetwork
local [ started ]
Runlevel: default
net.eth0 [ started ]
tor [ started ]
syslog-ng [ started ]
consolekit [ started ]
autofs [ started ]
netmount [ started ]
gpm [ started ]
xdm [ started ]
alsasound [ started ]
ntp-client [ started ]
vixie-cron [ started ]
ntpd [ crashed ]
privoxy [ started ]
sshd [ started ]
local [ started ]
Runlevel: sysinit
udev-mount [ started ]
devfs [ started ]
dmesg [ started ]
sysfs [ started ]
udev [ started ]
Runlevel: single
Runlevel: shutdown
savecache [ stopped ]
killprocs [ stopped ]
mount-ro [ stopped ]
Runlevel: boot
hwclock [ started ]
sysctl [ started ]
hostname [ started ]
modules [ started ]
fsck [ started ]
root [ started ]
mtab [ started ]
localmount [ started ]
bootmisc [ started ]
net.lo [ started ]
urandom [ started ]
termencoding [ started ]
keymaps [ started ]
swapfiles [ started ]
procfs [ started ]
tmpfiles.setup [ started ]
consolefont [ started ]
Dynamic Runlevel: hotplugged
Dynamic Runlevel: needed
dbus [ started ]
Dynamic Runlevel: manual
B) /etc/fstab has the following entry
/dev/md0p1 /home ext3
C) The second Raid partition /dev/md0p2 is not mounted
as part of the boot process, which is done intentionally so. |
|
Back to top |
|
|
Metuchen n00b
Joined: 14 Dec 2010 Posts: 4
|
Posted: Mon Jan 14, 2013 2:13 am Post subject: try partprobe |
|
|
You may want to try inserting a "partprobe /dev/md0" into your localmount script somewhere. It's a part of the "parted" ebuild. |
|
Back to top |
|
|
Goverp Veteran
Joined: 07 Mar 2007 Posts: 1967
|
Posted: Mon Jan 14, 2013 8:13 am Post subject: |
|
|
Quote: | After some update of my Gentoo installation, mounting /home does not work any more
and I do not understand what is going wrong because I did not change neither hardware
not Linux configuration. |
Could you share some information that update, and what levels of udev and openrc you were and now are using? It might be worth posting your dmesg output to pastebin as well.
FWIW, I've been running a partitioned RAID 5 array for the last 2 or 3 years with no problem. _________________ Greybeard |
|
Back to top |
|
|
Rainer06 n00b
Joined: 25 Mar 2006 Posts: 13 Location: Nuremberg
|
Posted: Sun Jan 20, 2013 4:58 pm Post subject: |
|
|
Thank you for the replies. The recommendation to use partprobe in the localmount script was resolving the problem.
The reason why the problem occured now is that local is now called at a later time in the runscript.
In the local startup, I had some parameter changes for the RAID array which were having the same
effect as partprobe.
Unfortunately, I do not know the previous versions of udev and openrc any more, but the new and different
sequence for local as part of the runscripts was definitely causing this problem. |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54033 Location: 56N 3W
|
Posted: Sun Jan 20, 2013 6:28 pm Post subject: |
|
|
Rainer06,
emerge genlop
will tell you when you compiled what versions of udev and how long each build took.
It depends on /var/log/emerge.log for input data _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
|