how can I fix: 0516-1147 lv 'may be partially mirrored' ?
Hans Vestman
(My problems began when a firmware upgrade failed on a 7025-F50
at a time when the OS also had an upgrade - from 4.2.1 to 4.3.3)
lsvg -l rootvg
:
:
0516-1147 lsvg: Warning - logical volume hd2 may be partially mirrored.
hd2 jfs 65 126 3 open/syncd /usr
:
lslv -l hd2
0516-022 lslv: Illegal parameter or structure value.
0516-304 lslv: Unable to find device id 0000000000000000 in the Device
Configuration Database.
hd2:/usr
PV COPIES IN BAND DISTRIBUTION
hdisk1 065:000:000 33% 013:000:022:018:012
hdisk0 061:000:000 36% 020:000:022:011:008
0000000000000000 004:000:000 ? ?
Would be enough to follow this procedure:
- deactivate rootvg, using varyoffvg
- exportvg
- importvg
- varyonvg
Or do I need to delete the mirror on hdisk0 first ?
Since the filesystem affected is /usr - do I then need to fix the problem on
site - in maintenance mode?
Or is it possible to solve the issue from a telnet session ?
Hope anyone has an easy way to fix my problem. The system in production is
working okay at the moment.
BR,
Hans Vestman
mike
would be especially nice.
Hans Vestman
Any other good ideas out there ... :-( ..?
...yes, we do have a mksysb ...
Hans Vestman
"mike" <m...@fundy.net> wrote in message news:3B09065C...@fundy.net...
Hajo Ehlers
your problem looks like a corrupted ODM.
I asume that HDISK0 and HDISK1 are your only rootvg disk.
From that information you must patch the odm to get the 00000.... back
to hdisk0
>PV COPIES IN BAND DISTRIBUTION
>hdisk1 065:000:000 33% 013:000:022:018:012
>hdisk0 061:000:000 36% 020:000:022:011:008
>0000000000000000 004:000:000 ? ?
So please try the following IRL
Let us know how its worked.
You may also read the following redbook:
AIX Storage Management
Document Number GG24-4484-00
found at www.redbooks.ibm.com
regards
Hajo Ehlers
Ram Pandiri
recreating it. "lslv -m hd2" can actually show you where the holes are,
but looks like simply four partitions have mirror copy missing. Run
"rmlvcopy hd2 1 hdisk0" and it will remove the partial copy. You can
them remake the copy - "mklvcopy hd2 2 hdisk0" (if you have 65 free
partitions on this disk then it will be successful).
Varyoff/export/import/varyon will not change anything ...
Ram
--
______________________________________________
Ram Pandiri pan...@austin.ibm.com
My views are mine, not IBMs.
Ram Pandiri
It does not look like ODM corruption issue ... LVM prints this infamous
all zeros for PVID if the mirror copy for some partitions do not exist
at all, a partial mirror case. Other than a corruption of VGDA, this
can happen if a third disk existed in the rootvg which housed the
partial copy of the hd2 and was later removed from the rootvg.
Ram
--
Hajo Ehlers
if you remove a disk from a system without the proper step there will be
information left in the odm.
So from a system point of view your ODM is not correct -> corrupted ;-)
regards
Hajo
Ram Pandiri
corruption. I was thinking in the lines of "obliterating some good
information with corrupt data" ... Anyway, in the current case, all the
info about the "possible" third disk is completely removed (also from
ODM, hence all the zeros).
Norman Levin
>
> I need a recommendation to the problem below:
> (My problems began when a firmware upgrade failed on a 7025-F50
> at a time when the OS also had an upgrade - from 4.2.1 to 4.3.3)
>
> lsvg -l rootvg
> :
> :
> 0516-1147 lsvg: Warning - logical volume hd2 may be partially mirrored.
> hd2 jfs 65 126 3 open/syncd /usr
> :
>
> lslv -l hd2
> 0516-022 lslv: Illegal parameter or structure value.
> 0516-304 lslv: Unable to find device id 0000000000000000 in the Device
> Configuration Database.
> hd2:/usr
> PV COPIES IN BAND DISTRIBUTION
> hdisk1 065:000:000 33% 013:000:022:018:012
> hdisk0 061:000:000 36% 020:000:022:011:008
> 0000000000000000 004:000:000 ? ?
>
> Would be enough to follow this procedure:
> - deactivate rootvg, using varyoffvg
> - exportvg
> - importvg
> - varyonvg
>
> Or do I need to delete the mirror on hdisk0 first ?
Then remove a copy for the hdisk? that has the smallest number of partitions - since
you said they are not completely mirrored. Usually the unmirror will get
rid of the flakey mirrored partition that is giving you the bogus bottom
line. Check again with lslv -m and if ok - just redo the mklvcopy
>
> Since the filesystem affected is /usr - do I then need to fix the problem on
> site - in maintenance mode?
> Or is it possible to solve the issue from a telnet session ?
>
> Hope anyone has an easy way to fix my problem. The system in production is
> working okay at the moment.
>
> BR,
> Hans Vestman
--
Norman Levin
Norman Levin
>
> Hi Hans,
> your problem looks like a corrupted ODM.
** no - I used to be able to recreate this consistantly by migrating an
lv from one disk to another. In that situation, you should only have a
single copy of the lv. Migrate moves a partition to the other disk by
calling mirroring code. For a short time, n-1 paritions have a single
copy and the nth parition exists on both disks. Definitely a strange
situation. Of course, since this is a migrate, unmirror is called and migrate
does this again for n times. If you interrupt the process, you can end up
with this situation.
--
Norman Levin
Hans Vestman
"Ram Pandiri" <pan...@austin.ibm.com> wrote in message
> Run
> "rmlvcopy hd2 1 hdisk0" and it will remove the partial copy. You can
> them remake the copy - "mklvcopy hd2 2 hdisk0"
Removing the partial copy - and remaking the copy fix'ed my problem!
Not enough disk space could have caused the problem. We now have 0 (zero)
PPs left ...! (so my customer has been instructed to add more disks!)
Thanks, again.
Hans Vestman