I think its the same reason I do it in my kickstarts. If a filesystem fills up it's not going to kill the OS. I have customers fill up /opt(a filesystem) with oracle logs. This causes oracle to have issues but doesnt impact the rest of the OS.
If I had the same issue..one option is to just symbolic link the directory without messing with LVM resizing. -Jim -----Original Message----- From: cobbler-boun...@lists.fedorahosted.org on behalf of Dick Davies Sent: Wed 7/4/2012 8:30 AM To: cobbler mailing list Subject: Re: [cobbler] Using 'lvresize' vs. 'lvextend' to correct space shortage in /var/www/cobbler Think it's fundamentally an EL 6.x issue where they changed the autopart defaults to split out into multiple filesystems. Still can't see why. On 4 July 2012 13:54, Bob Cochran <bcochra...@verizon.net> wrote: > On 7/4/12 2:59 AM, Bram Mertens wrote: >> >> On Tue, Jul 3, 2012 at 10:03 PM, Bob Cochran<bcochra...@verizon.net> >> wrote: >>> >>> I ran out of space for my cobbler repos in my CentOS 6.2 home for >>> cobbler. >>> Specifically, the default CentOS install allocates only 50 GiB of space >>> for >>> the LVM filesystem /dev/mapper/*/lv_root, which is mounted on /, and I >>> needed more of that. Since /opt comes under this, it was as empty of >>> space >>> as /var. So this meant using lvm commands to resize *lv_root and add >>> space >>> to it. I got into trouble with this unfortunately. >>> >>> Stuart suggested using 'lvextend' and 'lgextend' and perhaps I should >>> have >>> researched that more. Instead, I found this how-to and followed it: >>> >>> http://www.how2centos.com/centos-lvm-resizing-guide/ >>> >>> I robbed 1.2T of space from /dev/vg_cobbler1/lv_home using 'lvresize' and >>> got these messages: >>> >>> lvresize -L -1.2T /dev/vg_cobbler1/lv_home >>> rounding up size to full physical 1.20 TiB >>> .. >>> reducing logical volume lv_home to 577.98 GiB >>> >>> Then tried putting this space into /dev/vg_cobbler1/lv_root: >>> >>> lvresize -L +1.2T /dev/vg_cobbler1/lv_root >>> rounding up size to full physical extent 1.20 TiB >>> Extending logical volume lv_root to 1.25 TiB >>> Insufficient free space: 314573 extents needed but only 314572 available >>> >>> >>> what is this trying to tell me? And how do I fix it? Have I ruined the >>> filesystem(s) at this point? >>> >>> Should I have used 'lvextend' as suggested by Stuart, instead? >>> >>> The story will continue with what I did after this, but I would like to >>> stop >>> here and see what those of you more expert than me think. >>> >>> Thanks >>> >>> Bob >>> _______________________________________________ >>> cobbler mailing list >>> cobbler@lists.fedorahosted.org >>> https://fedorahosted.org/mailman/listinfo/cobbler >> >> Bob, >> >> At first glance the problem of following that particular guide is that >> it describes how to resize a SWAP partition. The difference between a >> SWAP partition and your home directory is that swap doesn't need a >> file system so resizing the partition is all that is needed. >> >> But your home partition contains a filesystem. >> >> To exted you need to resize a partition then resize the filesystem, to >> reduce you need to reduce the filesystem *first* then resize the >> partition carefully so the filesystem still fits on the reduced >> partition. >> >> Now if you did not make any other changes chances are you can still >> recover from this. You're system appears to be in an invalid state >> witht eh filesystem larger than the partition. But as far ar I know >> it can still work. >> >> First of all backup your data before you make any more changes! >> This is really important so I'll repeat: create a backup and verify it >> on a different system. >> >> Then I'd suggest to resize the partition back to it's original size. >> Then reduce the size of the filesystem first and then reduce the size >> of the partition again. >> >> The basic instructions are available at >> http://tldp.org/HOWTO/LVM-HOWTO/reducelv.html but it is very brief. A >> longer guide is available at >> http://www.tcpdump.com/kb/os/linux/lvm-resizing-guide/shrink.html. >> >> Good luck >> >> Bram > > > Bram and Ramon, > > Thank you very much for your help with this. I understand my mistakes better > now. I do have a complete backup of my CentOS system: I cloned the entire > hard drive to a spare hard drive, and then did all my resizing work from the > spare drive. So the original system drive is untouched. > > It is extremely rare for me to need to do LVM resizes. I think I've done > only one before, a few years ago. After making the above mistakes yesterday > (and actually making them a lot worse by using resize2fs on *lv_root) I > decided to simply reinstall CentOS 6.2 on that hard drive, reinstall cobbler > and its dependencies, and get the partition sizing correct from the start. > > I think the Cobbler installation instructions need to contain a note that on > CentOS systems which are using LVM, the root partition isn't going to get > enough space in a default installation, so an LVM resize will be needed or > even a reinstall of CentOS. I didn't understand this clearly until James > pointed it out to me -- that on my system /var had run out of space. It is a > lot easier to get the OS partitioned correctly before one is well into using > cobbler. > > Since I do have the original system drive, I will practice an LVM resize on > that. I can see myself needing to work with LVM much more in the future. > > Thanks! > > > Bob > > _______________________________________________ > cobbler mailing list > cobbler@lists.fedorahosted.org > https://fedorahosted.org/mailman/listinfo/cobbler _______________________________________________ cobbler mailing list cobbler@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/cobbler
_______________________________________________ cobbler mailing list cobbler@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/cobbler
