Raid 5 Reshape does not speed up

  • Hallo Freunde,


    ich versuche mein RAID 5 von 3x3tb auf 4x3TB, 3x4TB zu erweitern (das es dabei 3TB nicht nutzbaren Speicher gibt ist mir klar, aber die 4TB Festplatten waren günstiger als die 3TB Festplatten). Ich habe deshalb mit dem Grow Button die neuen Festplatten hinzugefügt. Leider ist der Prozess extrem langsam und braucht so circa einen Monat. Ich habe bereits


    root@frederiknas:~# sysctl dev.raid.speed_limit_max

    dev.raid.speed_limit_max = 5000000

    und

    root@frederiknas:~# sysctl dev.raid.speed_limit_min

    dev.raid.speed_limit_min = 500000


    cat /proc/mdstat liefert


    Code
    Personalities : [raid6] [raid5] [raid4] [linear] [multipath] [raid0] [raid1] [raid10]
    md127 : active raid5 sdj[7] sdh[6] sdg[5] sdf[4] sdb[1] sda[3] sdc[2]
          5860530176 blocks super 1.2 level 5, 512k chunk, algorithm 2 [7/7] [UUUUUUU]
          [=>...................]  reshape =  8.9% (263264660/2930265088) finish=26706.2min speed=1664K/sec

    Egal wie ich speed_limit_min und speed_limit_max verändere, die Geschwindigkeit des Reshape verändert sich nicht. Die Prozessor- und Arbeitsspeicher Auslastung ist dabei die ganze Zeit gering. Hat jemand einen Tipp, wie ich die Geschwindigkeit des Reshape erhöhen kann?


    LG

    F





    Hello friends,



    I am trying to expand my RAID 5 from 3x3tb to 4x3TB, 3x4TB (that there is 3TB of unusable storage in the process is clear to me, but the 4TB disks were cheaper than the 3TB disks). I therefore used the Grow button to add the new hard drives. Unfortunately, the process is extremely slow and takes about a month. I already have



    root@frederiknas:~# sysctl dev.raid.speed_limit_max

    dev.raid.speed_limit_max = 5000000


    and


    root@frederiknas:~# sysctl dev.raid.speed_limit_min

    dev.raid.speed_limit_min = 500000



    cat /proc/mdstat provides


    Code
    Personalities : [raid6] [raid5] [raid4] [linear] [multipath] [raid0] [raid1] [raid10]
    
    md127 : active raid5 sdj[7] sdh[6] sdg[5] sdf[4] sdb[1] sda[3] sdc[2]
    
    5860530176 blocks super 1.2 level 5, 512k chunk, algorithm 2 [7/7] [UUUUUU]
    
    [=>...................] reshape = 8.9% (263264660/2930265088) finish=26706.2min speed=1664K/sec



    No matter how I change speed_limit_min and speed_limit_max, the reshape speed does not change. The processor and memory usage is low all the time. Does anyone have a tip on how I can increase the speed of the reshape?



    LG


    F

    • Offizieller Beitrag

    Unfortunately, the process is extremely slow and takes about a month

    :/=O

    I don't know if this makes much sense. You would end up much sooner by creating a new Raid from a backup. The problem could be cutting the service during the process.

    If you do, I'd consider switching to ZFS if you really need a raid.

    Sorry, I don't know how to help you with your real problem, I don't use mdadm. Maybe geaves can help you.

    • Offizieller Beitrag

    Maybe geaves can help you

    Nope, I've read this twice and if I'm reading this correctly there are 3x3TB in a existing Raid5, to grow the array 1x3TB and 3x4TB have been added, nothing like putting an existing array under stress and to cap it off there probably isn't a backup.


    But to answer the OP's question take a look at this thread so in a nutshell can this be speeded up -> No!!

  • Hi there, thanks for your help! Sorry for my late reply.



    chente why do you think its a bad idea? I thought a lot about the best way of using these disks and considered this as the best solution. SnapRaid does not support the size and number of disks of the raid (as far as I know). Pooling two separate raid5 would be kinda laborious and I would lose another disk of storage. Is it even possible to stop the running process without loosing data?


    geaves I would be happy to reach the max speed of the disks but i don't think this is the bottleneck. In the linked thread Cyprus125 reached speeds of 16813K/sec which is about 10 times faster than mine. Do you have any ideas what could be the problem?

    My backup is stored in another town and it would take me more than a day to fetch it, so I really want to avoid this.


    Thanks for your help again.



    LG
    F

    • Offizieller Beitrag

    In the linked thread Cyprus125 reached speeds of 16813K/sec which is about 10 times faster than mine

    Yes but he's using Enterprise drives, big difference, have you looked at the site he links too

    Do you have any ideas what could be the problem

    Adding 4 drives at once to an existing array, however I did find this in my watched threads which at the time I found interesting, look on page 2 around post 26, for that OP that could have gone either way

    • Offizieller Beitrag

    SnapRaid does not support the size and number of disks of the raid (as far as I know).

    Why not? SnapRaid supports many more disks than in your setup.

    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.1.4 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4


    omv-extras.org plugins source code and issue tracker - github - changelogs


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

  • Hey thanks for your fast reply.


    For the last two hours Ive been messing with stuff but could not achieve an increase of speed. I think the linked thread could work because Ive tried everything else and it seems to have worked for others but i dont dare to try, so I guess I will just wait till its done.


    Regarding to snapraid i think that i misunderstood a few things......... If I visit my backups 'host' next time I will take it with me and do a snapraid.


    LG

    F

Jetzt mitmachen!

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