0

Im searching for a way to get the most out of "spare" HDDs. Following setup given:

  • 1x 2 TB
  • 2x 3TB
  • 2x 6TB

What i want/need is:

  • Parity (1x HDD can fail. For example, one of the 6TB. Data should still be accessable)
  • Most storage possible, without "dead space"
  • Attach new HDDs (any size) to increase storage (optional)

For me, it dosnt matter what kind of technology/software is used. ZFS/RAID/lvm/< name here >

As described above, a single HDD failure should be tolerated (The biggest single attached HDD for example)

I tought about a "JBOD", but cant find a final answer about how jbod is implemented. Wiht or without striping. And the most important point is, that the parity part is missing...

Im not sure what happens in lvm, when a hdd is missing, i think the whole volume is unavailable.

Long story short, some magic is needed ;)

Thanks for any advice.

EDIT #1: mhddfs/unionfs While searching i found "mhddfs" & "unionfs": https://serverfault.com/a/543684/569434 mhddfs seems a good way, execpt the parity point.

Marc
  • 127
  • 1
  • 7
  • 1
    Sell small drives on eBay, use the money to buy larger drives. – Michael Hampton Mar 01 '21 at 18:31
  • Or i could just buy all hdds in one size, sure. But thats was not what i asking for. Sometimes its importanter to know how to achive something rather then the result itself. How would you solve this? – Marc Mar 01 '21 at 18:51

2 Answers2

0

raid 1 of the 2 6TB gives you 6 TB.. Split the 3 TB drives into 2+1. raid 5 of the 3 2TB partitions gives you 4 TB. Then raid 1 of the remaining gives you 1 TB. Total is 11 TB.

Partition all the drives so they are labeled. The 6 and 2 TB drives are one partition. The 3 TB drives are split into 2TB + 1 TB.

Use mdadm to set up the raid, create the mdadm.conf file and set up a monitoring job to tell you if a disk fails. Once that's done and tested, add all the storage to lvm and enjoy.

stark
  • 360
  • 1
  • 13
0

Partition your disks as follows:

  • 2 TB disk: 1 x 2 TB partition
  • 3 TB disks: 1 x 1 TB partition, 1 x 2 TB partition
  • 6 TB disks: 1 x 1 TB partition, 1 x 2 TB partition, 1 x 3 TB partition

It is critical that each of the 1 and 2 TB partitions are exactly the same size from disk to disk.

Create the following RAID partitions:

  • 1 x RAID-5 partition using all four 1 TB partitions (3.2 TB)
  • 1 x RAID-5 partition using all five 2 TB partitions (8 TB)
  • 1 x RAID-1 partition using both 3 TB partitions (3 TB)

This gives you over 14 TB of storage and should be fault tolerant to the loss of any one physical disk.

pmdba
  • 281
  • 1
  • 6
  • I dont have a 1TB drive. Only 2/3/6 TB. – Marc Mar 01 '21 at 12:12
  • Not sure how I missed that. I have updated the answer accordingly. – pmdba Mar 01 '21 at 12:19
  • And out of that 3 RAID volumes, create a linear LVM? What when i want to add a new 6TB HDD? – Marc Mar 01 '21 at 12:31
  • RAID arrays/configurations aren't something that you can easily manipulate like an LVM. Whatever you build, you will be stuck with until you can provide a similarly-sized alternative. If you *know* you're going to have another 6TB disk, then don't do anything until you have it; that would change the entire recommended partitioning/RAID scheme. – pmdba Mar 01 '21 at 12:39
  • Thank you very much. I will try it later out. Out of curiosity, do you think i can achive the same with btrfs? – Marc Mar 01 '21 at 14:22