Migrate 2 disk RAID1 to 3 disk RAID5 array.

  • Hello


    I have been using OpenMediaVault since the start of the project (and Volker's FreeNAS, for years before).


    I have OMV (v1.11) on an HP uServer (N40L).


    I upgraded from a single 3TB disk to 2x3TB disks configured as RAID1 (/sdb + /sdc as /md127).


    I have just added a new 3TB disk (/sdd) to increase capacity. I am having problems migrating from the 3TB RAID1 to what I understood would be a new 6TB array (RAID5?)


    After installing the new drive, I used the webGUI to add the device under 'RAID Management'. After several hours of re-building a degraded array, the output of 'mdadm --detail /dev/md127', is:



    I thought I next had to grow the RAID using something like 'mdadm --grow --raid-devices=3 /dev/md127'.


    I was not clear if command 'resize2fs' was then needed.


    However, the output of these commands is:

    Zitat

    root@OMV1:~# mdadm --grow --raid-devices=3 /dev/md127
    mdadm: /dev/md127: no change requested


    root@OMV1:~# resize2fs /dev/md127
    resize2fs 1.42.5 (29-Jul-2012)
    The filesystem is already 732566356 blocks long. Nothing to do!


    I have read several posts on this forum (and Adding storage drives and changing RAID setup to suit), but I don't know if the only method is to degrade the array.


    I have my data in a back-up, and so if it is easier, I can simply delete the 3TB RAID1 array, and rebuild a new 6TB RAID (I assume RAID5 is most appropriate?)


    However, I thought it would be possible to migrate the array.


    I would really appreciate any help with this...


    Regards


    Andy

  • ...In the end, I decided to remove the existing RAID1 array (3 disk!?!?) and create a new RAID5 using the OMV WebGUI.

    Progress is now positive... I have a RAID5 array building with 3x 3TB disks. This shows a capacity of 6000.92 GB (which I think is correct?)


    Array re-sync progress...


    However, now I do now have a question regarding 'chunk size'.
    I understand this parameter can tune performance based on array I/O. My data file-size is very mixed - the NAS is a general file store/media server.
    After research, I conclude a large chunk size for RAID5 is advised (...?)
    I built with mdadm defaults, and assume 512k is classed as 'large'.


    If anyone advises otherwise, I'll halt the array re-syncing and try to tune this parameter.


    Many Thanks

  • Maybe this can bring you some answers (and a headache^^) : http://alephnull.com/benchmarks/sata2009/chunksize.html


    To say the least, I'm not sure changing your chunk size on 3 disks RAID5 is really helpful, and above all if you only store media files without playing them, not many I/Os...


    But if you absolutely want to do it, this guys recommends a chunck size of 128KB.


    Greetings

    - ASROCK FM2A88X-ITX+ (SATAIII (6Gb/s) x6 (for the DATA), mSATA x1 (for the OS))
    - AMD A6 7400K 3.5GHz
    - Corsair 2Go DDR3 1333MHz C9 (x2)
    - Intel Corporation 82574L Gigabit Network Connection
    - COOLER MASTER G450M (80+ bronze)
    - WD Red 2To 64Mo 3.5" SATAIII (6Gb/s)
    - 32 Go SSD mSATA KingSpec Half-Size Solid State
    - Fractal Design Node 304 black (HDD 3.5" x6)
    - RAID 5 XFS
    - OMV 4.1.35-1 Arrakis

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!