Opened 12 years ago

Closed 12 years ago

#120 closed defect (fixed)

Mondorestore fails with kernel panic on opensuse 10.2

Reported by: RickFriedman Owned by: bruno
Priority: normal Milestone: 2.2.2
Component: mondo Version: 2.2.1
Severity: normal Keywords:
Cc:

Description

While Mondoarchive creates the backup with no problem, Mondorestore definitely has a problem trying to restore it.

Upon trying to boot the archive DVD, the boot process starts and a few seconds later terminates with the following messages:

No filesystem could mount root, tried: minix iso9660 Kernel Panic not syncing: VFS: unable to mount rootfs on unknown block (0,1)

Attachments (3)

mindi.log.tar.gz (5.4 KB) - added by RickFriedman 12 years ago.
compressed /var/log/mindi.log
lsmod_mindi.tar.gz (6.9 KB) - added by plwrenn 12 years ago.
lsmod and mindi.log
mindi_lsmod.tar.gz (6.9 KB) - added by plwrenn 12 years ago.
lsmod and mindi.log after patch

Download all attachments as: .zip

Change History (41)

comment:1 Changed 12 years ago by RickFriedman

Actually... a small typo in my entry.

The kernel panic message says: "unable to mount rootfs on unknown block (1,0)"

comment:2 Changed 12 years ago by bruno

  • Milestone changed from 2.2.1 to 2.2.2

Do you use the mindi-busybox package ?

comment:3 Changed 12 years ago by bruno

Could you try again with the latest versions available mondo 2.2.1 + mindi 1.2.1 + mindi-busybox 1.2.2 ?

comment:4 Changed 12 years ago by plwrenn

I am having the same problem as Rick. The following is a little information about my install:

OpenSuSE 10.2 x86_64 Kernel 2.6.18.2-34-default

I build my 64-bit rpms from the following suse 10.2 rpm.src files:

mindi-1.2.1-1.suse10.2.src.rpm mindi-busybox-1.2.2-1.suse10.2.src.rpm mondo-2.2.1-1.suse10.2.src.rpm

My first thought is that this problem may be isolated to 64-bit suse since I have 2 32-bit installs of opensuse 10.2 running Mondo perfectly. If I can help by providing any more information please let me know.

Thanks for all of you hard work Bruno. We all appreciate it.

comment:5 Changed 12 years ago by RickFriedman

I use the packages that were installed from the SuSE 10.2 distribution repository (http://download.opensuse.org/distribution/10.2/repo/oss). Those packages are:

mondo 2.2.0-14 mindi 1.2.0-11 busybox 1.2.1-20

I can find no package named mindi-busybox in any SuSE repository. Is there such a SuSE package?

BTW... the problem isn't isolated to 64-bit SuSE. My install is 32 bit.

comment:6 Changed 12 years ago by RickFriedman

I went to ftp.mondorescue.org/suse/10.2, downloaded and installed the following packages:

mindi-1.2.1-1.suse10.2.i586.rpm
mindi-busybox-1.2.2-1.suse10.2.i586.rpm
mondo-2.2.1-1.suse10.2.i586.rpm

I then did a test backup of a subdirectory in my ~ directory to DVD. Once the backup was done (with no problem) I rebooted and attempted to boot from the DVD. I entered "interactive" at the boot prompt. The boot started up but about 3 seconds into it halted with the same errors that I originally reported.

Rick

comment:7 Changed 12 years ago by bruno

  • Status changed from new to assigned

Could each of you provide the result of lsmod and attache the /var/log/mindi.log file please ?

comment:8 Changed 12 years ago by RickFriedman

Here is the output from my lsmod:

Module                  Size  Used by
r128                   41088  2 
drm                    71316  3 r128
xt_pkttype              6016  3 
ipt_LOG                10624  8 
xt_limit                6912  8 
pppoe                  21696  2 
pppox                   7816  1 pppoe
af_packet              29320  2 
ppp_generic            31380  6 pppoe,pppox
slhc                   10752  1 ppp_generic
snd_pcm_oss            53376  0 
snd_mixer_oss          21248  1 snd_pcm_oss
snd_seq_midi           13824  0 
snd_seq_midi_event     11520  1 snd_seq_midi
snd_seq                60272  2 snd_seq_midi,snd_seq_midi_event
binfmt_misc            15368  1 
button                 10896  0 
battery                14340  0 
ac                      9476  0 
ip6t_REJECT             9344  3 
xt_tcpudp               7296  4 
ipt_REJECT              9472  3 
xt_state                6272  13 
iptable_mangle          7040  0 
iptable_nat            11524  0 
ip_nat                 21804  1 iptable_nat
iptable_filter          7168  1 
ip6table_mangle         6528  0 
ip_conntrack           52960  3 xt_state,iptable_nat,ip_nat
nfnetlink              11032  2 ip_nat,ip_conntrack
ip_tables              17604  3 iptable_mangle,iptable_nat,iptable_filter
ip6table_filter         7040  1 
ip6_tables             18756  2 ip6table_mangle,ip6table_filter
x_tables               18180  10 xt_pkttype,ipt_LOG,xt_limit,ip6t_REJECT,xt_tcpudp,ipt_REJECT,xt_state,iptable_nat,ip_tables,ip6_tables
ipv6                  263584  27 ip6t_REJECT
apparmor               55572  0 
aamatch_pcre           18304  1 apparmor
loop                   20488  0 
dm_mod                 60184  0 
sg                     38044  0 
sd_mod                 24576  0 
usb_storage            82112  0 
scsi_mod              136712  3 sg,sd_mod,usb_storage
usblp                  17792  0 
snd_mpu401             12712  0 
snd_mpu401_uart        13056  1 snd_mpu401
snd_rawmidi            29824  2 snd_seq_midi,snd_mpu401_uart
ns558                   8960  0 
gameport               19464  2 ns558
shpchp                 40224  0 
ide_cd                 42272  0 
cdrom                  38432  1 ide_cd
snd_seq_device         12812  3 snd_seq_midi,snd_seq,snd_rawmidi
pci_hotplug            35136  1 shpchp
snd_intel8x0           36764  2 
snd_ac97_codec         95648  1 snd_intel8x0
snd_ac97_bus            6400  1 snd_ac97_codec
snd_pcm                86916  3 snd_pcm_oss,snd_intel8x0,snd_ac97_codec
snd_timer              27908  2 snd_seq,snd_pcm
sis_agp                12676  1 
snd                    61188  16 snd_pcm_oss,snd_mixer_oss,snd_seq_midi,snd_seq,snd_mpu401,snd_mpu401_uart,snd_rawmidi,snd_seq_device,snd_intel8x0,snd_ac97_codec,snd_pcm,snd_timer
soundcore              13792  1 snd
snd_page_alloc         14472  2 snd_intel8x0,snd_pcm
sis900                 27392  0 
mii                     9600  1 sis900
agpgart                35528  2 drm,sis_agp
ohci_hcd               23428  0 
usbcore               114896  3 usb_storage,usblp,ohci_hcd
i2c_sis96x              9604  0 
i2c_sis630             11532  0 
i2c_core               25216  2 i2c_sis96x,i2c_sis630
parport_pc             42084  1 
lp                     15652  0 
parport                39752  2 parport_pc,lp
ext3                  141192  2 
mbcache                12804  1 ext3
jbd                    70324  1 ext3
edd                    13892  0 
fan                     8964  0 
sis5513                17160  0 [permanent]
thermal                18568  0 
processor              34664  1 thermal
ide_disk               20480  4 
ide_core              129992  4 usb_storage,ide_cd,sis5513,ide_disk

I've compressed & attached the mindi.log from the test backup I did earlier.

Changed 12 years ago by RickFriedman

compressed /var/log/mindi.log

Changed 12 years ago by plwrenn

lsmod and mindi.log

comment:9 Changed 12 years ago by nadavkav

i have the same error on fedora core 6 with (it's a k12ltsp distro) mondo-2.2.0-2.k12ltsp.6.0.0 mindi-1.2.0-2.k12ltsp.6.0.0

comment:10 Changed 12 years ago by Bruno Cornec

>  i have the same error on fedora core 6 with (it's a k12ltsp distro)

Please install the latest 1.2.1 of mindi (and also 2.2.1 of mondo) and
give me back your /var/log/mindi.log file.

comment:11 Changed 12 years ago by bruno

Hummm looking at your lsmod result I guess there has been a change in ext3. Could you try the following patch on your /usr/sbin/mindi and report back again:

--- mindi/mindi (révision 1034)
+++ mindi/mindi (copie de travail)
@@ -67,7 +67,7 @@
 USB_MODS="usb-storage usb-ohci usb-uhci usbcore usb_storage hid uhci_hcd ehci_hcd uhci-hcd ehci-hcd usbkbd usbhid"
 CDROM_MODS="$TAPE_MODS $FLOPPY_MODS $IDE_MODS af_packet cdrom isocd isofs inflate_fs nls_iso8859-1 nls_cp437 sg sr_mod zlib_inflate $USB_MODS $PCMCIA_MODS"
 NET_MODS="sunrpc nfs nfs_acl lockd loop mii 3c59x e100 bcm5700 bnx2 e1000 eepro100 ne2k-pci tg3 pcnet32 8139cp 8139too 8390 vmxnet vmnet"
-EXTRA_MODS="$CDROM_MODS vfat fat loop md-mod linear raid0 raid1 xor raid5 lvm-mod dm-mod jfs xfs xfs_support pagebuf reiserfs ext2 ext3 minix nfs nfs_acl nfsd lockd sunrpc jbd"
+EXTRA_MODS="$CDROM_MODS vfat fat loop md-mod linear raid0 raid1 xor raid5 lvm-mod dm-mod jfs xfs xfs_support pagebuf reiserfs ext2 ext3 minix nfs nfs_acl nfsd lockd sunrpc jbd mbcache"

 LOGFILE=/var/log/mindi.log
 FDDEVICE=/dev/fd0             ; # 1.44MB floppy #0

comment:12 Changed 12 years ago by plwrenn

how should we go about patching the shell script? should we just comment out the existing lines from USB_MODS to +EXTRA_MODS and then add you changes?

comment:13 Changed 12 years ago by plwrenn

nevermind my last post. I used the patch command. thanks.

Changed 12 years ago by plwrenn

lsmod and mindi.log after patch

comment:14 Changed 12 years ago by bruno

Another point:

Could you check the FAQ Q20 and report back again ?

comment:15 Changed 12 years ago by plwrenn

I added ramdisk_blocksize=1024 to my boot command and then ran mondoarchive. I received the exact same error when I attempted to boot from the CD I created.

comment:16 Changed 12 years ago by bruno

I don't understand exactly wht you did. In fact you don't need to re-run mondoarchive. You should just boot the produced media with the additional ramdisk_blocksize=1024 parameter at your loot loader prompt. Is that what you did ?

comment:17 Changed 12 years ago by plwrenn

I booted with ramdisk_blocksize=1024 parameter at my boot loader prompt. But then after booting re-ran mondoarchive and created media. Then rebooted to the new media and still had the same error when booting from media.

comment:18 Changed 12 years ago by bruno

No you need to use that parameter when you boot the built image, not to boot your own system. So when the mondo DVD/CD gives you the isolinux prompt there you add ramdisk_blocksize=1024

comment:19 Changed 12 years ago by plwrenn

Ok, I tried that and received the same error as before.

Sorry about the confusion. Your instructions were clear. I just didn't read them carefully enough at first.

comment:20 Changed 12 years ago by raagamuffin

I received the same error on a Fedora Core 5 (release 18_2257). I believe it may be something to do with ext3.

A mondoarchive/rescue process with an ext2 filesystem (with the same data) works without any hitches.

HTH.

comment:21 Changed 12 years ago by RickFriedman

Bruno, I tried your suggestion as well. On booting the Mondoarchive DVD, I added the ramdisk_blocksize=1024 option to the boot.

Unfortunately, it seemed not to make a difference. The same kernel panic error occurred as before.

comment:22 Changed 12 years ago by raagamuffin

A Correction to my previous comment. It does NOT work with ext2. It works with a previous release of FC5 - 17_2187 & ext2. Sorry for any confusion caused.

comment:23 Changed 12 years ago by garfield

I do happen to have the same Problem, opensuse 10.2, x86_32, kernel 2.6.18.2-34-default, / mounted ext3 => mondoarchive works, same Error on boot (tried with DVD+RW AND CD-R)

Also on the same machine I have a Suse 10.0-System (/ on ReiserFS) A DVD+RW created on this system booted and compared just fine.

Latest Mondo (2.2.1-1) and Mindi (1.2.1-1) in both cases

comment:24 Changed 12 years ago by captainkfb

I have the same problem. Virgin SuSE 10.2 install with default ext3 formatting. Tried the patch and ramdisk_blocksize=1024, to no avail.

comment:25 Changed 12 years ago by bruno

Another idea:

Could you try editing your local mindi script and remove acpi=off from the boot line generated ?

That could also be a problem in your cases.

Bruno.

comment:26 Changed 12 years ago by garfield

@bruno:

Thanks for your tip (removing acpi=off). Did not work for me, though. :-(

comment:27 Changed 12 years ago by plwrenn

I also tried removing acpi=off and it did not help.

But just to make sure I did it right:

I edited line 18 of /usr/bin/mindi

From: ADDITIONAL_BOOT_PARAMS="acpi=off apm=off devfs=nomount noresume selinux=0 barrier=off" To: ADDITIONAL_BOOT_PARAMS="apm=off devfs=nomount noresume selinux=0 barrier=off"

comment:28 Changed 12 years ago by tomdenecke

I usually work with mkCDrec (also using busybox/isolinux), but there is the same problem regarding OpenSuSE 10.2.

When I checked out the differences between the SuSE 10.0/10.1 and 10.2 kernels, I saw that SuSE 10.2 does NOT have ext2 compiled into the kernel anymore.

Therefore the kernel could not mount the initrd with ext2. The only filesystem which is compiled into kernel is minix. Using minix as the filesystem for mounting initrd brought the success.

Maybe using initramfs is a better approach of booting a rescue system.

Hope it helps you guys for the mondorescue project.

Regards Tom

comment:29 Changed 12 years ago by captainkfb

Thanks for the tip Tom. What specific changes did you make to mount initrd with minix rather than ext2?

comment:30 Changed 12 years ago by nadavkav

i had the same issue with a fresh install of FC6 (it's a K12LTSP Distro) and loading the mondo rescue DVD with ramdisk_blocksize=1024 kernel param solved my panic issue :-)

thanks !!!

comment:31 Changed 12 years ago by bruno

Here is the patch:

--- /usr/sbin/mindi     2007-01-20 17:05:51.000000000 +0100
+++ mindi       2007-01-20 17:04:52.000000000 +0100
@@ -2485,10 +2485,16 @@
        echo -en "..."
        dd if=/dev/zero of=$tempfile bs=1k count=$ramdisk_size &> /dev/null || Die "Not enough room for temporary ramdisk
(TurnTgzIntoRdz)"
        echo -en "..."
-       mke2fs -b 1024 -m 1 -i 2048 -F $tempfile >> $LOGFILE 2>> $LOGFILE
+#
+#      Changed 2 lines to make mindi work on OpenSuse 10.2, which has no ext2 support buildin
+#      so we use minix for the initrd. Jacob Wiersma (jw) 20.01.2007
+#
+#      mke2fs -b 1024 -m 1 -i 2048 -F $tempfile >> $LOGFILE 2>> $LOGFILE
+       mkfs.minix  $tempfile >> $LOGFILE 2>> $LOGFILE
        echo -en "..."
        mkdir -p $mountpoint
-       mount -t ext2 -o loop $tempfile $mountpoint || Die "Cannot loopmount $tempfile to $mountpoint! The reason may be
missing support for loopfs or ext2 (or both) in the running kernel."
+#      mount -t ext2 -o loop $tempfile $mountpoint || Die "Cannot loopmount $tempfile to $mountpoint! The reason may be
missing support for loopfs or ext2 (or both) in the running kernel."
+       mount -t minix -o loop $tempfile $mountpoint || Die "Cannot loopmount $tempfile to $mountpoint! The reason may be
missing support for loopfs or ext2 (or both) in the running kernel."
        echo -en "..."
        old_pwd=`pwd`
        cd $mountpoint

comment:32 Changed 12 years ago by captainkfb

Thanks Tom and Bruno. Patch worked for me and my SuSE 10.2.

comment:33 Changed 12 years ago by wdef

Same issue with the Mindi test boot cd on Fedora Core 5: kernel panic following can't mount root with filesystem etc

While I'm not sure if the Mindi test cd is supposed to have any actual backup functionality (is it?), the above error was got rid of with noot options: compare ramdisk_blocksize=1024

Ran through to a busybox command prompt.

comment:34 Changed 12 years ago by billott

Patch fails for FC6 because util-linux for FC6 does not include mkfs.minix.

comment:35 Changed 12 years ago by garfield

after applying the patch it does work for me. Thanks.

comment:36 Changed 12 years ago by bruno

The patch proposed by Andree in rev [1099] should fix that issue for good. Could someone with a Fedora Distribution please check and report ? That patch has been in mindi on Debian for some time now with success.

comment:37 Changed 12 years ago by bruno

Concerning 2.2.1, the solution seems to apply the 2 following patches. Please report success or failure.

http://trac.mondorescue.org/changeset/1230?format=diff http://trac.mondorescue.org/changeset/1232?format=diff

(Positive feedback from Thomas Börkel and JeffS for VMWare)

comment:38 Changed 12 years ago by bruno

  • Resolution set to fixed
  • Status changed from assigned to closed
Note: See TracTickets for help on using tickets.