0

7 Server with 4 Discs and I have this configuration:

  • "/" In a LVM (Physical disk 1)
  • "/data" In LVM (Physical disk 1 + fake raid 0 of disks 2 and 3)
  • "/data1" Ext4 (Physical disk 4)

The server is a supermicro (motherboard model X8DTL) with 8gb of ram.

I need to upgrade to centos 7 because the dependencies in the newer software are only in this distro but I have afraid of fuck UP With the data in "/data"

How I can upgrade safely without screw with "/data"?

PS: I can't make a backup, the information is more than 5TB.

"/data" and "/data1" there are only standalone files (Text, spreadsheet files, multimedia files). The programs and associations are only in "/"

Edit: Here it is how the disks are arranged:

# lsblk -o NAME,FSTYPE,SIZE,MOUNTPOINT,LABEL
NAME                               FSTYPE              SIZE MOUNTPOINT LABEL
sda                                linux_raid_member 931,5G            GLaDOS:0
└─md0                              LVM2_member         1,8T
  └─vg_glados_media-lv_data (dm-3) ext4                3,6T /data
sdc                                linux_raid_member 931,5G            GLaDOS:0
└─md0                              LVM2_member         1,8T
  └─vg_glados_media-lv_data (dm-3) ext4                3,6T /data
sdb                                                    1,8T
├─sdb1                             ext4                500M /boot
├─sdb2                             LVM2_member        97,7G
│ ├─vg_glados-lv_root (dm-0)       ext4                 50G /
│ ├─vg_glados-lv_swap (dm-1)       swap                7,8G [SWAP]
│ └─vg_glados-lv_home (dm-2)       ext4               39,9G /home
└─sdb3                             LVM2_member         1,7T
  └─vg_glados_media-lv_data (dm-3) ext4                3,6T /data
sdd                                                  931,5G
└─sdd1                             ext4              931,5G /data1     /data1
sr0                                                   1024M
# df -H
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_glados-lv_root
                       53G   44G  6,6G  87% /
tmpfs                 4,2G   78k  4,2G   1% /dev/shm
/dev/sdb1             500M  132M  342M  28% /boot
/dev/mapper/vg_glados_media-lv_data
                      3,9T  3,7T   28G 100% /data
/dev/mapper/vg_glados-lv_home
                       42G  862M   39G   3% /home
/dev/sdd1             985G  359G  576G  39% /data1
Maxtrix
  • 495
  • 5
  • 14
  • Do you have extra space in your volume group? – Cristóbal Ganter May 14 '16 at 00:31
  • `/` has some space but `/data` is full. – Maxtrix May 14 '16 at 00:37
  • You can use `redhat-upgrade-tool`. If not installed, install it via `yum install redhat-upgrade-tool`. Run it as `redhat-upgrade-tool --network 7.0` assuming repo is standard. If you have your own repo or CD, then change `--network` to suit your infra. Here's the doc for [redhat-upgrade-tool](http://www.unix.com/man-page/centos/8/redhat-upgrade-tool/). If you prefer to use pre-upgrade assistant, check [this](https://wiki.centos.org/TipsAndTricks/CentOSUpgradeTool) out. – alvits May 14 '16 at 00:59

2 Answers2

0

You have two options:

1) upgrade the existing installation. You could follow this RHEL manual for for example.

2) Make a fresh install, but a) tell the anaconda that you wish to do partitioning manually and b) carefully pick the correct partitions to format and to install the OS.

The latter option is much risky than the former. Also, you will lost any history/credentials/etc and would need to configure everything again.

If you have some spare disks to make a backup of your /data partition then better do it in both cases

Serge
  • 6,088
  • 17
  • 27
  • I cant make a backup (there are more than 5TB of information) :( – Maxtrix May 14 '16 at 00:03
  • Then I would recommend you to build a `testbed` with the same fs structure but using a single hard drive (the size does no matter, just to model the situation), install there CentOS 6.7 and go through all steps for option 1) with this machine. When you are confident you could repeat this safely on your actual server, then upgrade it. – Serge May 14 '16 at 00:07
  • You even could play with a virtual machine (for example, kvm of virtualbox) so you need not any extra hardware at all to model the process – Serge May 14 '16 at 00:16
  • That is a good idea, I wil search if in virtualbox can you simulate raid arrays to make more accurate, – Maxtrix May 14 '16 at 01:04
  • I just simulated the scenario with virtualbox and reinstalled above "/" Centos 7. I works like a charm and I have the data secured. Thanks Next week I will try this in the real machine. – Maxtrix May 17 '16 at 22:11
  • it is really nice) – Serge May 17 '16 at 22:13
0

I think you should add a new disk to your volume group. With this new space you can create a new logical volume and there you can try new experimental installations without compromising the rest of the system.

Cristóbal Ganter
  • 119
  • 13