I walk in the office on a fine Sunday morning to find out electricity has gone down restarting all boxes. One of my Debians wouldn’t leave runlevel 1 because of a logical volume not being mapped. I started the a typical bottom up troubleshooting procedure and at the first step (forget the vgscan and lvm libraries loading) :
timon:~# pvdisplay
Couldn’t find device with uuid ‘CMXpz2-HQQw-RO4k-5K2J-0mO4-oxTY-WvkY7W’.
— Physical volume —
PV Name unknown device
VG Name data_vol_grp
PV Size 74.53 GB / not usable 0
Allocatable yes (but full)
PE Size (KByte) 4096
Total PE 19079
Free PE 0
Allocated PE 19079
PV UUID CMXpz2-HQQw-RO4k-5K2J-0mO4-oxTY-WvkY7W— Physical volume —
PV Name /dev/hdd
VG Name data_vol_grp
PV Size 37.27 GB / not usable 0
Allocatable yes (but full)
….
This was – of course – creepy, knowing the drive was brand new, seen by the BIOS and detectable with a simple `fdisk -l`! Also check out the contradiction in italic.
After an hour of playing around … given I know the device path all I had to do is a `pvdisplay /dev/hdc` to get a healthy listing right there :O
I carried on with a `vgchange -a y data_vol_grp` followed by a `fsck.reiserfs` and finally mounted the volume and all was good!
Just thought I’d save some time and ATP burning surprise fellow admins should you ever experience this, especially that (as sledomly as it happens) google wasn’t that helpful.
Happy Hacking
really !! wow !! cool !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
that was nice man .. next time write something we can understand !!
Talk about computers with split personality hehe..
You see, the point where id get a uuid contradiction would more or less be the point where i pack my things and leave hehe.. or just blame it on the manufacturer /other admin / OS vendor (if any), n call it a wrap..
la2 bas spot on dude, one more gotcha to hang on the wall *thumbs up*… speaking of which, id like us to have a bit of an LVM talk sometime if you dont mind..
peace
Good job Yaman!
Honestly I never did RAID (if LVM is actually some RAID thing, btw!) in Linux. I am quite ok with such things in Sun Disk Solstice and the good old Veritas Vol Manager.. What I like most about those tools is that you get support from the vendors if you really get stuck. Life tends to be simple: Search The Fancy Web? Yeah right! Pick up Your Phone and Dial is a panacea!
This reminds me of a problem we had once in one of the telecommunication registers running on a Solaris-based cluster. As we had to `newfs` the RAW drives (talking about co0d1t0 hehe) /dev/rdsk/c0t0d0 or Cotodo so to speak, a cluster ID mismatch came to life.. Of course, that will cause your machines to go to runlevel 1 (or S). Anyway, commands to change your runlevel are: init n, telinit n, and IM! Yes, 5 minutes on IM made me successfully bring up the cluster and finally reach /sbin/rc3!
Take care…
though I didn’t get e thing of what you wrote here, it seems that everybody is cheering for your great performance, great job abulyomon way to go….