OMV - mdadm --> Upgrade to Snapraid + AUFS (or else?) install problem

  • OMV - mdadm --> Upgrade to Smapraid + AUFS (or else?) install problem


    Hi!


    2 years ago i have built a small NAS.
    Based on asrock a330GC (2core 4 thread Atom + 2x512MB DDR2 RAM) 2 sata onboard and a PCI


    extension slot.
    I installed a 8 port sata card in it. (highpoint rocket raid 1820 ---> recognized flawlessly


    :) )
    Then I installed the OMV on a 40GB IDE notebook drive plugged in an 2,5-->3,5 IDE adapter. ( I


    wanted 10 sata port XD)
    My data drives was first 3x160GB in RAID5 mdadm and then I installed another one and grow it


    up.
    This setup was running without problem 1 year then I added a 300GB drive with ext4 (not in the


    array)
    6month back I removed from my workstation 2 big drive filled with data, formatted with ntfs ,


    try it out maybe OMV can recognize it. No problem at all.(except speed)
    This drives are a 3TB WD green and a 1,5 TB Samsung ecodrive.


    Here I'm now.
    I would like to move all my data drives in this nas.
    So the what I want.
    I want to remove all the small drives from it, (after backuping them).
    Then install 2x500Gb in RAID1 mdadm for work drive (3ds max, photoshop etc...)
    And I'd like to buy 3x3Tb drive and make a Snapraid from them, fill them with data and install


    the existing 3TB wd in it (All together 4 drive 1 parity (3x3TB WD RED new, 1 WD 3TB green


    old)


    I installed the new 1.9 IMV on a virtual machine to try out the snapraid.
    I'm quiet confused.


    I created 1 drive for the OMV . Installed. Then add 3 more drive .
    Added 1 to parity and to to content and data.
    Checked the pool option.
    I shared the Disk 1-2-3 (snapraid disks to see that's going on)
    1-2 is the data 3 is the parity.
    I palced som files on the 1'st drive and some on the second.
    Synced them.
    Now i see in the pool share DISK1 and DISK2. But what i thought is there is no subfolder here, just my files on the 2 drives.


    So I installed AUFS (half hour to find out it has new name.... :S)
    I added the Disk1 and Disk2 in the AUFS. Then shared it.
    And in the aufs folder I don't see nothing what's on my disk.


    I like to have 1 big volume for the movies, pictures etc.
    But if there is a MOVIES folder on DISK1 and another on the DISK2 - how can i browse both in 1 directory ? Is it possible ?
    How should I set it up?


    Peter

    • Offizieller Beitrag

    You have snapraid setup correctly but don't use the pool share if you are going to use aufs.
    You have aufs setup correctly.


    Don't share the individual drives. Share only the aufs pool. All writes should go to the pool. aufs has a hard time tracking writes to individual drives. You can make shared folders that are subfolders of the pool and share those if you want individual shares.

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


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


    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!

  • And what's happens if my motherboard fail and I have to move the pool into another mashine ?
    As I said, I don't see the files on the individual drives, so it hase to some way to the OMV recognize it.
    As far as I know, the OMV recognize mdadm arrays by itself, but what's with the pools, such as snapraid and aufs ?


    On the other hand, I reed in this forum that, the AUFS is not more developed (and not as stable as others), something called "white holes".
    Can you give me an idea what to use , or just stick with aufs, and hope it does not crash on me ?
    For example there is the Greyhole or the mhddfs (whitch is much more stable but more havy on the load.)
    You see my rig, what could be the best ? Should I upgrade the hardwer aswell ?

    • Offizieller Beitrag

    Moving the pool to another machine should be no problem.
    If you don't see files on the individual drives, there is probably a permissions issue.
    mdadm is raid and a filesystem is on top of it. snapraid and aufs are on top of the filesystem.


    snapraid calculates parity from files on the data drives and writes it to the parity drive. If you remove snapraid, it does not remove your files.
    aufs writes to the drive with the most free space and reads from all the branches.


    aufs is being replaced by overlayfs. I use aufs myself and have no problems. aufs can't "crash". The worst thing that can happen is it only writes to one drive.


    greyhole is more resource intensive because it is doing a lot more. It is backup, pooling, and redundancy. snapraid is just redundancy and aufs/mhddfs are just pooling. mhddfs is more resource intensive because it is a fuse filesystem.


    With your system, I would use aufs. I would switch to overlayfs when it becomes available.

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


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


    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!

  • Erlyer I tryed some out and there was some rules for example I can tell the pooling solution how to use my drives.
    For me It's not an issue if 1 drive is written, and when it's become's 90-95% switch to the next, but in AUFS I don't see any propertyes.


    So you say I should see the files on the individual drives too , just set the permissions?
    I have currently 1 user with read and write permission. How could it be wrong ?
    I set always the inherit ( inherit writing properties) becouse I will use 2 user.
    1 for me, (R/W)
    1 guest --> everybody else with just read permission.
    But now it's in a virtual mashine with just the RW user.


    When the overlayfs come out can I switch smoothly or should I format the drives ?

    • Offizieller Beitrag

    You keep assuming that aufs is going to fail. I haven't even been able to replicate aufs not working. It just seems to affect a few users.


    I don't know how you are looking at the individual drives to see if there are files on them. You will have to explain that more.


    No need to create another user.


    I don't have a timeframe for overlayfs. The switch should be smooth. All of these work on top of the filesystem. So, there will never be any need to format just because you are switching pooling methods.

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


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


    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!

  • I tell you how I did it, and maybe you can point the point where I made a mistake.


    -Installer Openmediavault, Extra plugins, Snapraid and unionfilesystem.
    Then I went to the physical disks-->file systems and formatted my 3 new virtual HDDs with ext4 .
    Then went to Snapraid and added the Disk C to the Drives list with Parity ticked.
    After than I added the DiskA and DiskB to the list with content and Data Ticked on.
    Then pressed sync and wait until tell "it's done, nothing to do".
    After then I created the AUFS pool with the DiskA and DiskB and went to Users, add Name, pass, pass confirm OK.
    Then created a Shared Folder from the AUFS pool and set it in the SMB too.
    Then I added the Disk A-B-C for the shared folder and set my user to DiskA-B-C and AUFS pool Read and Write with inherit operation ticked.


    And so If I move some stuff in the AUFS pool I don't see it on the Disk A-B-C share.
    What did I missed ?
    Should I check something in the ACL setting ? Or add my user onto a group ?
    I got no idea .

    • Offizieller Beitrag

    Go to the Support Info tab under System Information. Select System - Shared Folder and post the output here.

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


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


    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!


  • This is there.

  • I think I figoured out something.
    There is a huge delay between applying a shared folder and then when it's appear on the SMB (win machines)


    I set the Disk B share to / (before these was DiskB/. I dont"t now why there is a difference becouse I thought there both directing in the root of the ext4 root . Or Am I wrong ?
    Now I set all DiskA-B-C to / in the shared folders . But the files don't appear, maybe I should restart the machine ? Or run a sync ?

  • So I think I get it.
    I don''t know why, but this version is set a default destination to a share (when create) on the name of the share.(create a folder) But what I want to share the root and then just create the folders under a win machine . So I just set all the shares to / (directing by this them to the root)


    So there still some question in my head If you don't mind.
    Question - 1
    So say there is a MOVIES folder on AUFS pool filled with fims. Filled with the AUFS pooling
    Which is appears on the disk like
    DiskA-->MOVIES-->Movie01,03,05,07...
    DiskB-->MOVIES-->Movies02,04,06,08 ....
    So there is MOVIES folder on both drives. And when I move the data drives to another machine (say the current one is dying) and I recreate a Snapraid pool diff it (no diff between parity and data on disks becouse it was synced before move) And recreate an AUFS pool from DiskA and DiskB and create a share with / so directed in the root of the snapraid.
    Then i gona see 1 MOVIES folder filled with Mocie01,02,03,04,05......
    So it can collect them ? It's "merge combining" them ? Or what ?


    I'm currently trying it and as far as I can tell AUFS is working in 500MB chunk sizes .
    (I'm moving tutorial movies on the aufs pool witch contain flv files size between 20-150MB and I see them in 500MB chunks divided in the drives (01-02-03-... till it cirka 500MB then switch to disk2)


    Question-02
    Say I build this NAS and fill the Snap pool with aufs pooling with a serious amount of data say 1-2 TB on the 3x3Tb pool. And I want to sync it. So I log in into the browser, and click on the sync, but I have to go away and turn off that computer so shut the browser , Is the sync gonna still runing ? And if so how can I watch later the status of it, just on CLI witch SSH ? (Like Raid5 rebuild if I click on it it's says its recovering and when I log in next time (when it's still not done) I can click on the detail button to see the progress, the speed whatsoever.)
    Is this possible in SnapRAID or did I just missed that ?


    Question-02


    I red about these whiteouts.
    This is just happening if the underlying filesystem is read only (or the AUFS doesn"t have permission to write on it ?)
    Am I correct ?
    Should i check the AUFS permissions somewhere ?
    What about deleting ?
    Say so I move my stuff to the new nas with AUFS and there I'm gonna clearify the folder structure by moving deleting renaming etc.
    Is that writen to all the disks at real time ?



    Question-03


    Spindown.
    Now with mdadm I use this 3x160 and all my drives (1,5TB and 3Tb ) with 10min spindown.
    Which is healthyer for the drives ?


    To use it or not ? Say I'm watching a file just 1-2 times from that pool (for example a movie at night.) When I'm browsing the media in for example a XBMC klient (Kodi) pointed on the AUFS pool, all my drives gonna spin up becouse of the MOVIES folder is spread on the drives.
    But when I choose my film say movie 056 whatsoever then all my unused disks (all exept which is containing my file) gonna spin down Am i right ?


    For work I gonna use the 2 500GB drive (WD Black talked about earlyer) with mdadm and try to set the spindown to say 1 hour or so.
    And I tend to use a 1,5 TB drive for torrent (spin all the time) sometimes cleaned and moved the content to the Snapraid.


    Is it a useable logical scenario ??


    Or should I just use a smaller drive example a 300GB older for torrent and add the 1,5 TB to the snapraid branch.
    Do you use spindown on computer which is running 24/7 ?


    If I use windows Network Folder and brownse my current mdadm raid then when I open the Windows Computer (I don't know whats the name im english sorry this is where you browse your drives on win)all my drives on this mdadm is spin up...
    what is not good for them becouse I just wanna spin then up when I use them.


    I guess when I open it windows is sending a "DIR" commant to the RIAD array for checking the aviable space and that's causing my problem. For me it's no problem to use the Network but my family is just not able to understand the "magic" of it.
    For them I used this solution to browse to MOVIES fodler. I hope I don't bother you with all my silly questions, but If I do something I think the best to do it properly not just try and try and try and fail. It wiser to ask :) Isn't ?:D


    Question-99


    Do you use Plex or XBMC? I tryed plex but it was make a lot of junk on my drives .

  • And 1 question more :)


    Can I defragment the drives ? If yes how ?
    I won't use win defrag tool's I guess those not ment to use to this purpouse.
    Do OMV or Snapraid has this option ?

  • between applying a shared folder


    and then when it's appear on the SMB (win machines)


    That should be below 10 seconds. I would even say they should show up within a second if you refresh.


    It's "merge combining" them ?


    Yes, you technically make a JBOD out of them. They look as one, but AuFS distributes the files over all disks in the pool.


    AUFS is working in 500MB chunk sizes


    I highly doubt that the chung size is that big, but I haven't checked myself.


    Is the sync gonna still runing ?


    Yes.


    And if so how can I watch later the status of it, just on CLI witch SSH ?


    Not sure...


    Is that writen to all the disks at real time ?


    AuFS is real-time.


    Which is healthyer for the drives ?


    Which Religion is better? There are believers of both setups, with and without spindown. I would't say one or the other is better for the drives lifespan.


    all my drives gonna spin up


    Not sure if this is caused by the Kodi Feature to wake up the NAS.


    gonna spin down Am i right ?


    They should.


    Windows Computer (I don't know whats the name im english sorry this is where you browse your drives on win)all my drives on this mdadm is spin up...


    Don't access them as Network Drives. Period.


    For me it's no problem to use the Network but my family is just not able to understand the "magic" of it.


    Put a shortcut on the Desktop.


    Can I defragment the drives ? If yes how ?


    You can, but you don't need to.


    Do OMV or Snapraid has this option ?


    Snapraid wouldn't have anything to do with it. And no, OpenMediaVault has no option either.


    Greetings
    David

    "Well... lately this forum has become support for everything except omv" [...] "And is like someone is banning Google from their browsers"


    Only two things are infinite, the universe and human stupidity, and I'm not sure about the former.

    Upload Logfile via WebGUI/CLI
    #openmediavault on freenode IRC | German & English | GMT+1
    Absolutely no Support via PM!

  • David,


    First: Thank's for the response.


    The desktop shourtcut is an insanly simple solution... I should've figured it out :D
    Then I've learned a lot of how this things work's in the background.
    So I think I've a solution.


    I got my ext 4 formatted drives and the snapraid, and here I have several options.


    1- I use AUFS to poole them together into 1 large pool until and so I can fill them with movies.
    2- when I want to clean up the folder structure, I switch the AUFS pool to an MHDDFS (becouse I don't want to use those WH files insted of realy delete and move, rename them.
    I know there gonna be a botleneck here (maybe I will use for this an AMD 965BE with 8gig undeclocked and undervolted as muscle.)
    3- For media purpouse I could use just the Snapraid Pool option as a read-only pool so I don't ruin my stiff and clean folder structure.


    The question what's remain is that how XBMC and plex works.
    I'm gonna read it about, and when I got question I'm asking :)


    If you have any idea, just spit it out :)


    The idea is to use some nicer looking movie browsing solution on or off the server.
    I guess the on nas solution is better, no question. The question is the setup.


    I don't want to ruin my file structure and fill them with a lot of junk.
    if plex download a cover or someting (I.E. cover.jpg) that's Ok but not else.
    I don't want to have other files on this array if have to then in another dedicated folder for junk.
    And if I delete a file I what would plex do ? Delete it from the plex browser or it have to be done manualy ?


    And what is this transcoding ?:D I'm quiet interested in all option, how to use efficiently my NAS.
    So I have a lot to learn and a lot to read, now I'm just getting to scratch the surface, how to use it.
    I'm going to read about and If I have someting I come back.
    I wrote this to be useful maybe for others. (I hope it no prob) Sorry maybe for the long diary's :D

  • 3- For media purpouse I could use just the Snapraid Pool option as a read-only pool so I don't ruin my stiff and clean folder structure.


    ??? What do you mean by stiff and clean folder structure?


    The question what's remain is that how XBMC and plex works.


    ? What do you wanne know?


    If you have any idea, just spit it out


    ?


    I guess the on nas solution is better, no question. The question is the setup.


    Plex is Server Side Database, like MediaBrowser is. XBMC can do it in both ways with various different setups. The standard is local Database on each MediaPlayer, but you can easily Setup a shared MySQL Database. Media Information (Artwork etc.) will be stored on each respective Media Player (and it is not suggested to share them anymore).


    I don't want to ruin my file structure and fill them with a lot of junk.


    I know what you mean. Though, it doesn't hurt you if there are nfo files inside your Movie Folders, which I highly suggest you to use (a Folder named for the respective Movie, then your Structure doesnt get messed up). But I have to say, I don't think Plex exports nfos or Artwork by default... so the question is, are you sure that Plex filled your Movie Folder with that stuff?


    And what is this transcoding ?


    Some devices can't play specific codes, or just profiles of a codec. If thats the case Plex can do transcoding which is a live conversation of the file to the requesting device.


    Greetings
    David

    "Well... lately this forum has become support for everything except omv" [...] "And is like someone is banning Google from their browsers"


    Only two things are infinite, the universe and human stupidity, and I'm not sure about the former.

    Upload Logfile via WebGUI/CLI
    #openmediavault on freenode IRC | German & English | GMT+1
    Absolutely no Support via PM!

  • By Stiff and clean I mean there is no WH files all over the Disks.
    When I move stuff on the NAS with AUFS there will be. I don't want it so when I orginize things I intend ot use MHDDFS


    Now on a VM I get 4-7 MB/s. And I don't see where is the boleneck. I guess the VM is, becouse the cpu is 5-6% the kernel load 0.4 (4core amd phenom is the host of the VM with 8 gig ram) ant the VM got 2gig ram and 95% max process allowed.


    Plex souns fine and I don't have any problem with nfo files, most torrent come with them a let them just there it's no prob the problem is when WH files appears.
    For filling the drives AUFS is betteg no question. But for organizing MHDDFS is way better to get decent result on the undeneath filesystem.
    I try to create a future proof array, where when I have to move a drive I dont wanna have problems with the WH-files or other.
    Now I trying out the MHDDFS in VM and then I will try the plex server.
    Can an XBMC clien use the plex database ? And don't heard about MediaBrowser, until now I just used the network to browse and open my files. I tried Plex way way back and probabaly I messed up something with the setup ( then I had to do all in CLI )
    Can you tell me your soution for pooling ? AUFS vs MHDDFS or else ?

  • By Stiff and clean I mean there is no WH files all over the Disks.


    And why do you care about that? They don't fill up much space and I think you can ignore them when sharing via CIFS/SMB.


    And I don't see where is the boleneck.


    mhddfs is.


    Can an XBMC clien use the plex database ?


    With PlexBMC, yes.


    Can you tell me your soution for pooling ? AUFS vs MHDDFS


    I don't use pooling at the moment. But I consider to do it in the future...


    or else ?


    3-Ware 9650SE-16ML.


    Greetings
    David

    "Well... lately this forum has become support for everything except omv" [...] "And is like someone is banning Google from their browsers"


    Only two things are infinite, the universe and human stupidity, and I'm not sure about the former.

    Upload Logfile via WebGUI/CLI
    #openmediavault on freenode IRC | German & English | GMT+1
    Absolutely no Support via PM!

    • Offizieller Beitrag

    The chunk size isn't 500 MB. aufs re-evaluates which drive has the most free space every 30 seconds (approximately).

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


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


    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!

  • Now I tried the same snapraid with AUFS on the same VM and 18-20MB/sec. Deffinietly the VM is the botleneck. With native run I think I would get around 75-80 witn AUFS .
    The hardwer cannot be the limit in this preticulat situation.


    Can Openmediavault support XBMC somehow ? I mean on database level.

    • Offizieller Beitrag

    Install mysql plugin.

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


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


    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!

Jetzt mitmachen!

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