If you are explicitly specifying the volume group path like this:
        /dev/VG/LV

Then it doesn't need the links that are created by that bit of code.
/dev/VG/LV will always be there as long as the volume group is available
on the system.  The mapper links (/dev/mapper/VG-LV) are what get
created by the problem code.

So, with your lv output:
  LV       VG     Attr   LSize  Origin Snap%  Move Log Copy% 
  crashlv  rootvg -wi-ao  8.00G                              
  locallv  rootvg -wi-ao  1.00G                              
  projects rootvg -wi-ao 20.00G                           
  rootlv   rootvg -wi-ao  5.00G                              
  swaplv   rootvg -wi-ao  2.00G                              
  varlv    rootvg -wi-ao  3.00G                              

/dev/rootvg/crashlv should always exist, regardless of whether or not
the mapper is running / configured.  The kdump init script creates links
in /dev/mapper/.

Maarten Broekman
Email: [EMAIL PROTECTED]


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Jan Frode Myklebust
Sent: Monday, August 04, 2008 4:15 AM
To: [email protected]
Subject: [rhelv5-list] Re: RHEL6 Wishlist -- Less Regressions, more QA

On 2008-08-01, Broekman, Maarten <[EMAIL PROTECTED]> wrote:
> Correct.  With the awk patch, both /dev/VG/LV and /dev/mapper/VG-LV
will
> work.  Without the awk patch, only /dev/VG/LV will work since
> /dev/mapper/VG-LV are just links to /dev/VG/LV.  Later in the init
> script, it checks the DUMPDEV which is /dev/mapper/VG-LV on your
system.
> Since the link wasn't correctly made because of the use of cut instead
> of awk, that DUMPDEV doesn't exist.
>
> The use of awk is definitely far more appropriate.  I was just noting
> that kdump would work fine with /dev/VG/LV as the dump device instead
of
> /dev/mapper/VG-LV.
>

I don't understand why DUMPDEV=/dev/VG/LV should work any better..
would be great if it did, as that would be a simple workaround for
this problem, but the VGRP will still be empty because of the use of
"sed", so it shouldn't be able to create the /dev/VG directory, or
/dev/VG/LV symlink..

Here's the interesting part of mkdumprd:

1524 if [ -n "$vg_list" ]; then
1525     emit "echo Scanning logical volumes"
1526     emit "lvm vgscan --ignorelockingfailure --mknodes"
1527     emit "echo Activating logical volumes"
1528     emit "lvm vgchange -a y --ignorelockingfailure"
1529     emit "DM_NUM=0"
1530     emit "for i in \`lvm lvs | awk '{ if (NR > 1) print \$1}'\`"
1531     emit "do"
1532     emit "   VGRP=\`lvm lvs | grep \$i | cut -d\" \" -f4\`"
1533     emit "   mkdir -p /dev/\$VGRP"
1534     emit "   if [ ! -e /dev/\$VGRP/\$i ]"
1535     emit "   then"
1536     emit "      ln -s /dev/mapper/\$VGRP-\$i /dev/\$VGRP/\$i"
1537     emit "      ln -s /dev/mapper/\$VGRP-\$i /dev/dm-\$DM_NUM"
1538     emit "      DM_NUM=\`echo \$DM_NUM 1 + p | dc\`"
1539     emit "      if [ -z \"\$noresume\" ]"
1540     emit "      then"
1541     emit "        /sbin/dmsetup.static resume
/dev/mapper/\$VGRP-\$i"
1542     emit "      fi"
1543     emit "   fi"
1544     emit "done"
1545 fi


Or am I missing something ? 


  -jf

_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list


_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list

Reply via email to