0

Regarding LVM Partition

I am using a Linux server machine (IBM) for a genome assembly task. The OS is Ubuntu, version 22.04.2 LTS (Jammy Jellyfish)

The machine has a memory capacity of 248 GB. There are 8 SAS hard disks installed in the system. You may take a look at the picture of the physical disks. Each physical disk has a capacity of 300 GB. A set of four physical disks is configured with a RAID5 partition. The RAID 5 partition details can be found [here] and **[here].

So, now there are two virtual drives - Drive 0 and Drive 1.

Drive 0 has two partitions: /dev/sda1 and /dev/sda2.

Details of /dev/sda1:<------click the link

Details of /dev/sda2:<------click the link

Drive 1 has two partitions: /dev/sdb1 and /dev/sdb2.

Details of /dev/sdb1:<------click the link

Details of /dev/sdb2:<------click the link

Here are the contents of the /etc/fstab.

Background:

I have been assembling the genome using Linked reads (378 million reads), using Universal Sequencing Technology’s TELLYSIS pipeline. While the assembly was progressing, I was caught off guard, seeing that there was no space left in the /dev/sdb2 partition. Consulting with the company, I was instructed that there must be at least 1 TB of free space before starting the assembly. Now I am worried because my assembly project is still halted.

There are 12 SAS disk slots available, of which 8 are currently in use. I can't add additional disks due to a lack of connectivity on the motherboard. One solution was to replace an existing drive with a SAS disk of higher storage capacity. I contacted a local vendor, and the quotation that they sent was for a SAS disk with a capacity of 1.2 TB, but the price is too high to afford at the moment.

What I am looking for:

Recently, I learned that using LVM, physical disks can be merged, and space can be extended by creating volume groups and logical volumes.

However, I am only a programmer and not experienced in handling Linux admin-level tasks. Nonetheless, I have decided to give it a try. The following is the step-by-step method I have decided to perform this task. (Note: I have already created a backup for all my data).

  1. /dev/sda2 has about ~800 GB of free space. Booting from a Linux bootable drive, I can delete this partition and later create a new primary partition to convert it to the "Linux LVM" type.
  2. Perform the same on the /dev/sdb2 (~700 GB of free space) partition.
  3. Create a volume group using these two partitions.
  4. Create a logical volume partition of size > 1 TB within this volume group.
  5. Mount it.

Do you think this is possible? Can you provide me with feedback on whether my solution is inadequate, potentially risky, or totally impossible? If it is impossible, can you suggest a method that I can afford to follow?

0

1 Answer 1

1

What you can do is to:

  1. Create PVs on /dev/sda2 and /dev/sdb2

    pvcreate -f /dev/sda2 /dev/sdb2

  2. Create VG from PVs above

    vgcreate vg_name /dev/sda2 /dev/sdb2

  3. Create LV in this VG

    lvcreate -L 1024GB -n lv_name vg_name

  4. Create filesystem on this LV

  5. Mount is using usual instruments

But be warned if any of the partitions /dev/sda2 /dev/sdb2 get broken you loose entire LV

2
  • Actually, the existing hard disks are quite old (~10 y). Also, what do you mean when you say "partitions get broken"? How likely is that to happen? Did you mean HDDs will be irreversibly damaged? Commented Sep 4, 2023 at 11:16
  • 1
    @VijithKumarV, what I mean is RAID5 can fail (failed 2 disks) and you loose all the info on this volume. And if these disks have 10 years working time you can expect fail in any time soon. Commented Sep 4, 2023 at 11:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .