The 2TB Storage limiation

[Aaron_Pritchard]

Hi Guys.

Running through the background briefly: Runing R77.30 MDS environment. The kernel limits Checkpoint OS to builds of 2TB on open servers. I have an MLM that is filling up and so have built a larger storage open server with a 12 TB partition to hold logs (and a root partition or 80GB, whch hosts the OS).

OK. this is fine.

Next question.

1. What is the best way to generate new symbolic links, to forward logs from /var/log/ to my new mounted drive '/var/log/logstore' ? While thinking of this answer, this is an MLM.... which means there are to be 3 CLMs attached. When our firewall team load SmartLog, it needs to be able to load seemlessly opening the correct mount with the logs from that CMA.

cli output below for reference.

[Mike_A]

Why use the symbolic link? I have always got past the 2TB limitation by, before install, killing all arrays in the RAID controller that will not hold the OS. Leave your 1x array at say 80GB, in this instance, and do not join your other disks to an array on the controller yet. Install the OS on your 80GB partition.

Once the OS is loaded, go back in and create your new array, RAID (whatever you want), and boot into OS. Once booted into OS you can use pvcreate, vgextend and lvextend to add your new disk space from the second array to the logical volume which is /dev/mapper/vg_splat-lv_log this will allow /var/log to surpass the 2TB limitation. sk114115 references exactly what im talking about. 

Since the disk space you are showing in your screen shot is already presented to the OS, did you modify /etc/fstab and mount  /dev/cciss/c0d1 to /var/log/logstore? If so then I believe you can remove this reference from /etc/fstab and follow the SK above to extend your LV on /var/log and not muddy the waters with the symbolic link. 

[Aaron_Pritchard]

Mike Andretta wrote:

Why use the symbolic link? I have always got past the 2TB limitation by, before install, killing all arrays in the RAID controller that will not hold the OS. Leave your 1x array at say 80GB, in this instance, and do not join your other disks to an array on the controller yet. Install the OS on your 80GB partition.

Once the OS is loaded, go back in and create your new array, RAID (whatever you want), and boot into OS.

^^^ correct. this is what i usualy do too, and have done to get to this stage.

Mike Andretta wrote:

Since the disk space you are showing in your screen shot is already presented to the OS, did you modify /etc/fstab and mount  /dev/cciss/c0d1 to /var/log/logstore?

^^^ yes i did. (i followed an sk, dont have it to hand.)

OK, i will try your other recommendations. thanks for that sk. i wasnt aware it would be possible to get around that 2TB limit another way. let me try and ill update. cheers

[Aaron_Pritchard]

hmmm, i seem to have a unknown source of 200gb.

[Expert@UKMLM01:0]# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_splat-lv_current
                       97G   13G   79G  15% /
/dev/cciss/c0d0p1     289M   24M  251M   9% /boot
tmpfs                 3.9G     0  3.9G   0% /dev/shm
/dev/mapper/vg_splat-lv_log
                       97G  4.5G   88G   5% /var/log
/dev/cciss/c0d1        11T  159M   11T   1% /var/log/logstore
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]# pvs
  PV                VG       Fmt  Attr PSize   PFree
  /dev/cciss/c0d0p3 vg_splat lvm2 a-   400.84G 200.84G

After then removing the custom fstab config, and rebooting, the 'logstore' mount is gone.

but, i dont see it now as 'free space'/ and ht eprevious 200gb is still showing

[Expert@UKMLM01:0]# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_splat-lv_current
                       97G   13G   79G  15% /
/dev/cciss/c0d0p1     289M   24M  251M   9% /boot
tmpfs                 3.9G     0  3.9G   0% /dev/shm
/dev/mapper/vg_splat-lv_log
                       97G  4.6G   88G   5% /var/log
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]#
[Expert@UKMLM01:0]# pvs
  PV                VG       Fmt  Attr PSize   PFree
  /dev/cciss/c0d0p3 vg_splat lvm2 a-   400.84G 200.84G

[Expert@UKMLM01:0]# fdisk -l

Disk /dev/cciss/c0d0: 450.0 GB, 450064605184 bytes
255 heads, 63 sectors/track, 54717 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

           Device Boot      Start         End      Blocks   Id  System
/dev/cciss/c0d0p1   *           1          38      305203+  83  Linux
/dev/cciss/c0d0p2              39        2387    18868342+  82  Linux swap / Solaris
/dev/cciss/c0d0p3            2388       54717   420340725   8e  Linux LVM

Disk /dev/cciss/c0d1: 12002.2 GB, 12002236260352 bytes
255 heads, 32 sectors/track, 2872777 cylinders
Units = cylinders of 8160 * 512 = 4177920 bytes

Disk /dev/cciss/c0d1 doesn't contain a valid partition table

[Mike_A]

It is right here, no?

Disk /dev/cciss/c0d1: 12002.2 GB, 12002236260352 bytes
255 heads, 32 sectors/track, 2872777 cylinders
Units = cylinders of 8160 * 512 = 4177920 bytes

Disk /dev/cciss/c0d1 doesn't contain a valid partition table