Increase the Logical volume - root

[HP-UX Increasing the ro]

Hello !!
I have one HP-UX server.. Now its shows / root is occupied 100 % . Many users are using this server to access some database,Please help me out to increase the size without rebooting also not in single user mode, here i dont have Tape drive for backup etc.. etc...

Looking forward a solution for this problem..

Thanks in advance,
Arun

[Jeeshan]

Hi Arun

take a look at this link

http://www.symmetricwebsites.com/

[Shrikant Lavhate]

Hi,

extending root lvol without reboot i.e. on the fly is quite risky. Again you dont have a tape for backup also. I suggest that even if you dont have tape make sure that you have backup on some network location to start with in case of failures.
Extending root file sys can be acheived using lvextend.
Methods using lvextend or online JFS can be found within forum search.

[Shrikant Lavhate]

Also, before you apply anything try to free up some space by trimming log files, removing unwanted softwares & temp files etc.

[MarkSyder]

It seems highly likely that you have a process writing to / that should be writing to a dedicated filesystem. Check this before you do anything drastic like increasing /.

Also, check /dev for files type f: a common mistake is to mistype a tape device name and produce a large text file (although that seems unlikely in your case as you say you don't have a tape drive).

Mark Syder (like the drink but spelt different)

[Pete Randall]

Whoa, folks!

First off, it is nearly impossible to extend your root volume. Secondly, to do so without a reboot or a tape device is literally impossible. Because /stand, swap and / need to be contiguous, a simple lvextend is not going to accomplish anything. The only feasible way to accomplish this it to restore from an Ignite make_tape_recovery backup, re-sizing your root logical volumes in the process. Since you have no tape device, that seems unlikely.

I would start with Mark's advice and invest my time in looking to find what is filling your root volume, rather than trying to accomplish the basically impossible task of enlarging. Use the find command to search /dev, as Mark indicated (-type f). Then use it to search / for core files and large files in general ( -name and -size).

Under normal circumstances, / should be almost completely static and there should be no reason for growth. You need to do some investigation into why it filled up and not waste your time trying to enlarge something that can't readily be enlarged.


Pete

[HP-UX Increasing the ro]

Thanks for all..

I have done with all basic and some troubleshoot to increase the volume in /.. and tried to delete some files under / but all contains applications many users
working on it like 24/7.... still now we all are able to work on it.. I hope after some times later i wount get the login prompt .. due to /dev - root 100% full.. it would be great if someone help me out with this issue..

[mobidyc · Accepted solution]

Hello,

i've wrote an article (in french) for extend root volume if you have a miror, you can found it here:
http://mobidyc.net/?id=23

please, adapt these commands in your situation.

basically, here are the commands passed in my case:
1/ broke the miror:
for i in /dev/vg00/lv* ; do lvreduce -m 0 $i ; done

2/ identify the PV with all its free extends by :
vgdisplay -v vg00 |tail -n 20

3/ pvmove lvol4 from used pv to free pv seen above :
pvmove -n /dev/vg00/lvol4 /dev/dsk/c1t2d0 /dev/dsk/c2t2d0

4/ after that, you can extend root :
lvextend -L 2048 /dev/vg00/lvol3

5/ you can re-pvmove the lvol4 :
pvmove -n /dev/vg00/lvol4 /dev/dsk/c2t2d0 /dev/dsk/c1t2d0

6/ redo your miror :
for i in /dev/vg00/lv* ; do lvextend -m 1 $i ; done

on another terminal, you can monitore the progress with:
for i in /dev/vg00/lvol* ; do t=$(lvdisplay -v $i |awk '/dsk.*stale/,/^$/'|wc -l) ; echo ${i##*/} Stale: $t ; done

--
Regards,
Cedrick Gaillard

[Rita C Workman]

OMG.....greetings oh interestingly named Author - I'll call ya H-pux

Please look at what Pete has said. You can not on the fly increase root ( / ).
Clean up "safe" files you can clean up. Try running the following to clean up old levels of patches you will never remove anyway:

cleanup -c (1 or 2)

Then look under /dev and see if some newbie wrote a file out under /dev/rmt instead of writing it to a tape unit. I see this one more than I care to think about.

...AND if your users are writing things out 24/7 why don't they have disk and a seperate mountpoint to put it there! So your O/S wouldn't be getting hammered.

Rgrds,
Rita

...it is less painful when you plan for growth, then force it to fit with a hammer...

[Rita C Workman]

H-pux (Arun)

I just read "here i dont have Tape drive for backup etc...etc..."

You have one HPUX server in serious peril. Your data is extremely unsafe. Get a cheap dat tape drive and at least do a make_tape_recovery. How on earth are you backing up your data???

You need to address this with the $$ mgmt. Because when something happens (and it will) who do you think they will hold accountable first. you.

Best Wishes,
Rita