Aufs, how does it work?

  • Can't find much about AUFS but I have set it up now and from what I understand you create one poolshare, and this folder uses both HDDs. But now I can't create shared folders right? What if I want many users and not all to have access to everything in the poolshare?


    I also want a few folders to be backed up automatically. Can I create 2 partitions on 2 drives and use rSync to backup both partitions?


    Thank you!

    • Offizieller Beitrag

    aufs tries to write equal amounts of data to each disk. Generally, the pool share is what you share but in your case, that doesn't give enough separation. So, just make more shared folders in side the pool share.


    So, if pool share is /pool/
    Make a shared folder for group #1 /pool/group1/
    Make a shared folder for group #2 /pool/group2/
    and so on.


    Use the rsnapshot plugin to backup shares at whatever frequency you want.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | 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!

  • aufs tries to write equal amounts of data to each disk. Generally, the pool share is what you share but in your case, that doesn't give enough separation. So, just make more shared folders in side the pool share.


    So, if pool share is /pool/
    Make a shared folder for group #1 /pool/group1/
    Make a shared folder for group #2 /pool/group2/
    and so on.


    Use the rsnapshot plugin to backup shares at whatever frequency you want.


    Great! I will try that. I've also found something called mhddfs. What's the real difference between it and AUFS? From what I understand it keeps everything in better order. Not spreading everything out. Which do you recommend? Do they both work on 1.0?


    When setting up plugins as plex, how do I point media? To the pool or the disk?

    • Offizieller Beitrag

    Both pool disks. aufs tries to keep the same amount of free space on each drive in the pool. mhddfs writes to a disk until it is full and then moves to the next.


    Right now, plex puts its database on one disk. You tell it in the plex web interface where to look for media and you can specify the pool directory or each disk in the pool. In the next few days, plex will stored its database on a shared folder.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | 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!

  • Both pool disks. aufs tries to keep the same amount of free space on each drive in the pool. mhddfs writes to a disk until it is full and then moves to the next.


    Right now, plex puts its database on one disk. You tell it in the plex web interface where to look for media and you can specify the pool directory or each disk in the pool. In the next few days, plex will stored its database on a shared folder.


    Which do you recommend. It should be more simple with mhddfs right? As I don't have to setup the whole Pool in Plex to search for movies and tv series etc? That way I know it will put everything at the right place?


    Do I loose any functionality by going with mhddfs?


    What do you mean with " In the next few days, plex will stored its database on a shared folder. " ?

    • Offizieller Beitrag

    I use aufs because I like balanced drives but I don't think it really matters. Some people had problems with aufs not balancing the free space but I haven't. mhddfs is a simpler concept so it might be a bit more reliable. I say flip a coin :)


    Right now, you don't get to pick the directory the database is stored in. You just to get pick the drive. We have been switching the plugins over to use shared folder instead of just picking the drive.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | 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 use aufs because I like balanced drives but I don't think it really matters. Some people had problems with aufs not balancing the free space but I haven't. mhddfs is a simpler concept so it might be a bit more reliable. I say flip a coin :)


    Right now, you don't get to pick the directory the database is stored in. You just to get pick the drive. We have been switching the plugins over to use shared folder instead of just picking the drive.


    Great!
    I have one more thougt...
    I have 4 disk (+1 system disk).
    3x 2TB
    1x 750GB


    I want to make these 4 disk a big pool with aufs or mhddfs (by the way, is it easy to add more disks to the same pool later on with both aufs and mhddfs?).
    I also want to use rsnapshot to take backup of a folder that is placed in the pool. But is there a way to reserve space on the 750GB disk to be used for rsnapshot backup? Because if they are all in the pool the risk is that aufs/mhddfs places the original and the backup at the same disk?
    And maybe aufs/mhddfs will fill up the volume too, so there is no room for rsnapshot?


    Sorry for my many newbie questions, but I don't want to do anything wrong, and it's hard to configure once all my data is there (I don't want to experiment then).

  • ryecoaaron has done a great job with the plugin so adding extra drives to the aufs pool is very simple, not used mhddfs so do not know about that.


    I don't think you can reserve space in aufs but you can use multiple pools so you could have the 3x2tb in a pool and the 750gb just mounted normally. Another option would be to partition the 750gb drive to the size you want for your backup, you should (not tried this myself) be able to use this partition as backup space and the other partition can be added to a pool.

    [align=center]banner@1x.jpg
    What I am watching/watched on my Popcorn Hour media player
    omv 2.1.1 stoneburner | 64 bit | omvextrasorg 1.29
    MySQL (Percona[size=8]) | Newznab | Sonarr | CP | HP | NZBget | YAMJ | CUPS


  • joq3 - I use mhddfs and i really like it. For me, it was more "unbreakable" (i kept having poor luck with AUFS). In addition, i like how it writes the data. It fills one drive first, then once it hits the user specified threashold, it moves on to the next drive. This way not all the drives are spun up at once. In addition, i have a better idea of were my data lives.


    With that being said, you can add a new disk into the pool anytime. It will start being filled up after the other disks are full. I started without any pooling, then enabled mhddfs and added 3 disk without issue.


    Good Luck!


  • joq3 - I use mhddfs and i really like it. For me, it was more "unbreakable" (i kept having poor luck with AUFS). In addition, i like how it writes the data. It fills one drive first, then once it hits the user specified threashold, it moves on to the next drive. This way not all the drives are spun up at once. In addition, i have a better idea of were my data lives.


    With that being said, you can add a new disk into the pool anytime. It will start being filled up after the other disks are full. I started without any pooling, then enabled mhddfs and added 3 disk without issue.


    Good Luck!


    I was decided for AUFS a moment ago, as it seems to be the most popular, most used pool-plugin. But now I'm unsure again!
    What kind of problems did you have with AUFS?


    And if I leave MFS unchecked it does what mhddfs does right? Fill one disk first?

  • What kind of problems did you have with AUFS?


    For some people the balancing did not work correctly.


    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!

  • For some people the balancing did not work correctly.


    Greetings
    David


    Sorry for keep asking questions, but I guess this will be helpful for other than me.
    Is this the latest mhddfs? http://omv-extras.org/debian/p…/o/openmediavault-mhddfs/


    From what I understand there are only benefits with mhddfs over aufs?
    Or is there a good reason to keep disks balanced?
    mhddfs pro's is that disks doesn't spin up when they aren't needed, but with aufs splitted files they will?


    Is there a performance advantage?
    Same thing with MFS, should it be checked or not?


    Are both aufs and mhddfs as well written and stable?


    Let's make this clear, please recommend what I should use :)

    • Offizieller Beitrag

    That is the latest mhddfs.


    Some people like aufs better due to the balance. With mhddfs, if you don't have a backup and one drive fails, you lose more files. You definitely won't spin up the other drives as much using mhddfs which may make them last longer


    Neither mhddfs nor aufs split a single file across drives. So, performance will always be equal to one drive with both.


    I wrote both plugins and the mhddfs plugin is almost identical to the aufs plugin.


    I literally just flipped a coin and it said use mhddfs :)

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | 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!


  • Great reply, thank you!


    mhddfs is as supported as aufs I guess? I think it will be fine, as I plan using rsnapshot to backup the important folders to another drive. And mhddfs should be the better choice as you say it will probably make the drives last longer, plus save energy.


    Is there a simple way to uninstall aufs and install mhddfs, don't want it to collide with eachother, so that aufs is removed completely.
    Havn't transfered my files yet so I can format and partition if needed!


    And what should the upper threashold be? Should I fill the drive up completely before moving to the next?

    • Offizieller Beitrag

    I support both plugins so that is the same. Not sure about kernel level support but they are both in wheezy. So, they should be supported the same.


    Just uninstall the aufs plugin and install the mhddfs plugin. They won't collide with each if you leave aufs though.


    I would set the threshold to the same as your largest file (probably 8gb unless you have some bluray rips). If you have fresh drives, install the mhddfs plugin and transfer your files to it using the pool shared folder. It will fill up the first drive until it hits the threshold and then start filling up the second drive and so on.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | 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!

  • hi ryecoaaron i had a little question about mhddfs too.
    I have successfully installed the plugin by downloading the deb file
    this is the configuration I did:



    Unfortunately I can not figure out how to configure the pool as a samba share. Which shared folder should I put in the field to bind? This is only a dummy shared? It can be 1 of the 3 branch?


    Thx

  • Even though I started using mhddfs I went back to aufs. With mhddfs I got much lower write-speeds. 60 MB/s instead of 110 MB/s which aufs gave me.
    So I'm back with aufs.


    I left MFS unchecked, and it doesn't seem to fill up my first drive before going to the next. Check this screenshot:

  • Weap,
    This is how i setup my pool. Below you can see D1, D2, and D3 were branches and poolshare was the bind share.





  • -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
    First of all, let me thank the developer of AUFS for this great program!


    It really helps to keep your Data balanced and organized.


    But I got a problem with plex, when I'm using AUFS.


    I got 4*4tb drives and thei are all Brances (1-4).
    The Bind Share is on the first Drive.


    When i want to configure PLEX int he web client, it doesn't show folders in the Bind Share folder, in the pooled Folder and even not in the NFS exported folder (but I don't know if i got the NFS export options right fpr this, bis it works with 2 Openelec machines and one FTV).


    After googleing i found something about the read and writes rights on this folder, but i don't got it.


    Maybe someone can help?


    (I'm running 1.0.2 Kralizec and the 3.2.0. backport kernal)


    Greetings,
    derbeisser
    -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------


    UPDATE: Everything is fixe'd. Thx

Jetzt mitmachen!

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