Hello Friends ,

I have some queries on increasing sapce in "Raid1" partition on centos 6

I have 2 HDD (sda & sdb) with four partition (swap, /boot, / and /opt), all are configured on Raid1. The four raid1 partition (md0, md1, md2, md3)on each HDD are taken from four "part" of each HDD

[root@mail ~]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 3.7T 0 disk
├─sda1 8:1 0 32G 0 part
│ └─md0 9:0 0 32G 0 raid1 [SWAP]
├─sda2 8:2 0 512M 0 part
│ └─md1 9:1 0 511.4M 0 raid1 /boot
├─sda3 8:3 0 100G 0 part
│ └─md2 9:2 0 100G 0 raid1 /
├─sda4 8:4 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt
└─sda5 8:5 0 1M 0 part
sdb 8:16 0 3.7T 0 disk
├─sdb1 8:17 0 32G 0 part
│ └─md0 9:0 0 32G 0 raid1 [SWAP]
├─sdb2 8:18 0 512M 0 part
│ └─md1 9:1 0 511.4M 0 raid1 /boot
├─sdb3 8:19 0 100G 0 part
│ └─md2 9:2 0 100G 0 raid1 /
├─sdb4 8:20 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt
└─sdb5 8:21 0 1M 0 part

I need to increase the space only on /opt , So i added one more HDD - sdc.

[root@mail ~]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 3.7T 0 disk
├─sda1 8:1 0 32G 0 part
│ └─md0 9:0 0 32G 0 raid1 [SWAP]
├─sda2 8:2 0 512M 0 part
│ └─md1 9:1 0 511.4M 0 raid1 /boot
├─sda3 8:3 0 100G 0 part
│ └─md2 9:2 0 100G 0 raid1 /
├─sda4 8:4 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt
└─sda5 8:5 0 1M 0 part
sdb 8:16 0 3.7T 0 disk
├─sdb1 8:17 0 32G 0 part
│ └─md0 9:0 0 32G 0 raid1 [SWAP]
├─sdb2 8:18 0 512M 0 part
│ └─md1 9:1 0 511.4M 0 raid1 /boot
├─sdb3 8:19 0 100G 0 part
│ └─md2 9:2 0 100G 0 raid1 /
├─sdb4 8:20 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt
└─sdb5 8:21 0 1M 0 part
sdc 8:32 0 5.5T 0 disk


Now my question is,

1. Is it possible to increase the space only on /opt ? (in this scenario im asking)
2. If possible, Can i do two equal partition on sdc and may i use it for sda4 & sdb4 to increase the space on md3 (/opt) ?

Hope my explanation is bit clear, please help on my query.

Thank you,

Since sdc would be a single disk it is a single point of failure so you would not get any RAID benefit by partitioning it and adding both partitions to a single RAID1 meta-disk (or to a filesystem no matter how you partition it).

If you're willing to take the risk of having no redundancy for /opt you could just use the entire sdc for a single partition and replace you're existing /opt with that. Alternatively you could make the sdc partition a subdirectory of /opt (e.g. /opt/myappdir) and move whatever needs the most space onto it while leaving everything else in your /opt.

Ideally you'd install yet another disk that you could pair with sdc for the new stuff.

Another way to get redundancy would be to reinstall using RAID5 for the 3 disks (assuming they're all the same size). You'd end up with the full space of 2 of the disks available which is what you'd have with a RAID1 with 2 disks and another disk not in RAID at all so the benefit of RAID5 would be having redundancy.

If you insist on abrogating the protection afforded by RAID1, why not just use RAID0 ?. I'm not sure mdadm will allow conversion this "on the fly".

Note, I am not advocating this, merely offerring it as another option.

@ MensaWater

>> Since sdc would be a single disk it is a single point of failure so you would not get any RAID benefit by partitioning it.
Im ok with this above statement. i ready to go forward

[root@mail ~]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 3.7T 0 disk
├─sda1 8:1 0 32G 0 part
│ └─md0 9:0 0 32G 0 raid1 [SWAP]
├─sda2 8:2 0 512M 0 part
│ └─md1 9:1 0 511.4M 0 raid1 /boot
├─sda3 8:3 0 100G 0 part
│ └─md2 9:2 0 100G 0 raid1 /
├─sda4 8:4 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt
└─sda5 8:5 0 1M 0 part
sdb 8:16 0 3.7T 0 disk
├─sdb1 8:17 0 32G 0 part
│ └─md0 9:0 0 32G 0 raid1 [SWAP]
├─sdb2 8:18 0 512M 0 part
│ └─md1 9:1 0 511.4M 0 raid1 /boot
├─sdb3 8:19 0 100G 0 part
│ └─md2 9:2 0 100G 0 raid1 /
├─sdb4 8:20 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt
└─sdb5 8:21 0 1M 0 part
sdc 8:32 0 5.5T 0 disk


We cannot change the set up in the above format , because it is a live mail server and in /opt zimbra (a mail server application) is running..

my doubt is , md3 is taken from sda4, sdb4. i need to increase the sapce on sda4, sdb4 to get incease the space in md3. Since sda4, sdb4 are- type "part" is it possible to increse the sapce ?

├─sda4 8:4 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt

├─sdb4 8:20 0 3.5T 0 part
│ └─md3 9:3 0 3.5T 0 raid1 /opt

if possible how can we do please tell me..

@ MensaWater

>>Another way to get redundancy would be to reinstall using RAID5 for the 3 disks (assuming they're all the same size). You'd end up with the full space of 2 of the disks available which is what you'd have with a RAID1 with 2 disks and another disk not in RAID at all so the benefit of RAID5 would be having redundancy.

MensaWater, here we are using the sofware RAID, Can you please tell me how to move my configuration from RAID1 to RAID5, Can i do this with out any data ?

Your problem is that your existing setup is RAID1. If you add another disk (or partition) it simply becomes another copy without giving any increase in space (i.e. it merely increases redundancy).

As 2 of us have noted using the 3rd disk alone (even if you mirror 2 partitions on it) as a separate meta-disk will leave you without any real redundancy. If you say you can't take down time even for planned maintenance then not having redundancy is a bad idea.

My suggestion, if you are willing to forego redundancy is to move the largest item off of md3 to the new disk.

In /opt you can run "du -sk * |sort -rn" to see which is the largest subdirectory.

You can then make a new subdirectory called something else (e.g. mkdir /opt/zimbra.new if /opt/zimbra is your largest) and mount your new disk or partition (e.g. /dev/sdc1) as /opt/zimbra.new.

You'd be able to copy all the files from existing directory to the .new directory. You'd want to do final copy with any application using the existing directory shutdown. You can find what is using it with lsof (e.g. lsof /opt/zimbra).

You can then move the existing to .old (e.g. mv /opt/zimbra /opt/zimbra.old) and recreate the directory to be an empty mount point (e.g. mkdir /opt/zimbra).

You'd then set up your /etc/fstab to mount the device used (e.g. /dev/sdc1) as /opt/zimbra. You'd unmount the /opt/zimbra.new and remount as /opt/zimbra. You'd then remount the filesystem.

Again the above would allow you to have the additional space of the new device for whatever you moved but would NOT be redundant if it is a single drive (even if you use multiple partitions).

Thank you very much for the information, I clarify a lot. Thank you.