FSCK problem : Logredo failed

[swtbeer]

I've copied a logical volume from a volume group to another :

cplv -v vgnew lvname

Then I've changed the logical volume to point to the new jfslog in the new volume group

chfs -a dev=/dev/lvname -a log=/dev/loglv00 /mountpoint

Until this point everythin is OK, but when I pass a fsck to the new logical volume, I get the next error :

laertes:/# fsck -p /dev/fslv02
log redo processing for /dev/rfslv02
logredo: not a log file /dev/rfslv02
Failure replaying log: 0
/dev/rfslv02 (/bbe/e): logredo failed, fsck continuing
/dev/rfslv02 (/bbe/e): 2583 files 3449304 blocks 2842152 free

I think this isn't a serious problem , but I would like to be sure. Anybody knows what is hapenning?.

Thanks in advance.
jcia

 

[theycallmetim]

unmount ALL filesystems pointing to that log, run logform to clean it, then run fsck and see what happens. Could just be some corruption, but with that error, if you did have a system failure you would have some problems.

 

[sushveer]

Hi Swtbeer

One can have a common jfs log for many file systems and if performance matters one can also have individual jfs log/each file system..

I noticed from your message there is a wrong procedure followed by u after cplv.There is a conflict of old common jfs log and new jfs log for file system in question

The step by step approach to create jfs log new logical volume is as follows:I recommend this is the better:

step1: Create a new jfs log lv as
#mklv -t jfslog -y <lvname> <vgname> 1 <hdisk#>

step 2:formating log as
#/usr/sbin/logform /dev/lvname

step3: You can modify /etc/filesystems and LVCB as
#chfs -a log=/dev/lvname /mnt

step 4:unmount file system
#umount /dev/lvname
or umount /mnt

step 5: check the file system integrrity
#fsck -y /dev/lvname or fsck -y /mnt

step 6:mount the file system one by one in that specified vg group
#mount /mnt

step7:
verify by using #df -k



sushveer
IBM certified specialist-p-series AIX5L System Administration
AIX/SOLARIS/WEBSPHERE-MQ/TIVOLI Administrator

 

[SPguru]

Given what you entered:

cplv -v vgnew lvname

The LVM is going to create and name a new LV on the dest. VG.

Then you enter:

chfs -a dev=/dev/lvname -a log=/dev/loglv00 /mountpoint

But you cannot chfs lvname that you entered on cplv because it is not what you want.

Then you:

fsck -p /dev/fslv02

Is fslv02 the name given from cplv? If so, then fslv02 needs to be entered on chfs.

****************

Character I/O requests are performed by issuing a read or write request on /dev/rlvX. (just for info)

 

[swtbeer]

Hi all,

I didn't put real names in my note, so It could be a little confused.

These are the real names :

I have two VGs

laertes:/# lsvg
rootvg
volapp21

I have lv06 in rootvg :

laertes:/# lslv lv06
LOGICAL VOLUME: lv06 VOLUME GROUP: rootvg
LV IDENTIFIER: 0043364a753a15df.13 PERMISSION: read/write
VG STATE: active/complete LV STATE: opened/syncd
TYPE: jfs WRITE VERIFY: off
MAX LPs: 512 PP SIZE: 16 megabyte(s)
COPIES: 1 SCHED POLICY: parallel
LPs: 192 PPs: 192
STALE PPs: 0 BB POLICY: relocatable
INTER-POLICY: minimum RELOCATABLE: yes
INTRA-POLICY: middle UPPER BOUND: 32
MOUNT POINT: /bbe/em610 LABEL: /bbe/em610
MIRROR WRITE CONSISTENCY: on
EACH LP COPY ON A SEPARATE PV ?: yes

I have a jfslog (loglv00) in volapp21 :

laertes:/# lsvg -l volapp21
volapp21:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
loglv00 jfslog 1 1 1 open/syncd N/A
lv02 jfs 950 950 3 open/syncd /bbe/ecs600
lv04 jfs 83 83 1 open/syncd /bbe/patrol
lv05 jfs 150 150 1 open/syncd /backup_nc
lv07 jfs 567 567 3 open/syncd /usr/
I want to move lv06 from rootvg to volapp21.

laertes:/# umount /bbe/em610
laertes:/# cplv -v volapp21 lv06

It returns new LV name : /dev/fslv02

I change the jfslog to the one in volapp21.

laertes:/# chfs -a dev=/dev/fslv02 -a log=/dev/loglv00 /bbe/em610

Then when I do fsck a get the next error :

laertes:/# fsck -p /dev/fslv02
log redo processing for /dev/rfslv02
logredo: not a log file /dev/rfslv02
Failure replaying log: 0
/dev/rfslv02 (/bbe/e): logredo failed, fsck continuing
/dev/rfslv02 (/bbe/e): 2583 files 3449304 blocks 2842152 free

I think that the new jfslog isn't in effect when I do FSCK, and LVM is searching for a jfslog for that LV and cannot find it. It also could occurs because the new jfslog is in use and it isn't formatted. Any idea ?


swtbeer

 

[Marksmen]

Does your logfile "/dev/loglv00" exist in vg volapp21? If not, you will need to create a log file with something like:

mklv -t jfslog -y jfslog2 volapp21 1
logform /dev/jfslog2

where jfslog2 is my logfile name. You can then run chfs pointing the log device for the filesystem to the jfslog2 log.

 

[hirschaj]

IMHO, this is one of the situations where IBM's procedure is wrong. I don't think the jfslog would have any data relating to the filesystem since it has never been mounted yet. I would mount the FS and forget about running the fsck. You always have your data in the old LV that you can convert back to if needed.


Jim Hirschauer

http://www.aixexpert.com

 

[CrystalWizard]

The jfs log does, yes, have information about the filesystem...if nothing other than the filesystem it's going to be the log FOR.

you run FSCK if you can't mount the filesystem.
If you can mount it and the only error you get is that the log failed, not superblock problems or something simular, then just run logform on the jfs log. if that doesn't fix it, remove the loglv and make a new one.

Which IBM procedure do you think is wrong?

 

[hirschaj]

The IBM procedure for copying an LV from 1 VG to another is what I am referring to. If you actually have tried it you will find that it gives the same error everytime you try to run fcsk before trying to mount the FS for the first time in the new VG. The errors in question have nothing to do actually mounting the FS.


Jim Hirschauer

http://www.aixexpert.com

 

[kHz]

I have copied numerous LVs to different VGs and never had an error, so I don't know what you think is wrong in copying an LV from VG to VG.

 

[hirschaj]

Ok kHz, try to keep up. We are talking about running the fcsk before you mount the FS for the first time in the new VG giving an error message. Please carefully read the posts. Copying the LV from VG to VG works great.


Jim Hirschauer

http://www.aixexpert.com