How to Kill Greyhole

  • I really love the concept of greyhole, and I gave it a good, honest try, but I have discovered that it simply is not the best fit for me. The experience was quite shocking, by the end.


    A few SMB shares, 3 drives, greyhole, sabnzbd, transmission. All of this together with some mass file transfers to the greyhole shares and KABOOM!


    I don't know if the kernel panicked or what, but the NAS went down hard. I reset the server and watched it boot. Logged into the console and KABOOM!


    I repeated this a few times...


    Had to boot into recover mode and remove greyhole from the init.d/ dir. Then it booted fine.


    Snapraid and aufs is next on the menu, as soon as I rsync as much data off as I can.

  • ...not good...
    check mhddfs (same plugin as aufs) and do some internet research on the topic. A lot of people report better stability from the combination Snapraid+mhddfs.
    I'm using Snapraid and it works really well.
    FT

  • There is really only one media type that causes people to have the need to span drives. It is video. I use symlinks to bring my media in to a main data drive from the others. It is fast, simple and all to easy. Works great. I watch all the people that have issues with RAID and just shake my head.


    Example:
    Drive 1 (where main media folder resides)
    create media shared folder and share it with samba
    /media/uuid1/media
    connect to share via pc and create folders (e.g. Movies, TV, Music, Photos, etc...)
    /media/uuid1/media/Movies
    /media/uuid1/media/TV
    /media/uuid1/media/Music
    etc.....


    Drive 2
    create storage 2 shared folder and share it with Samba (use default drop down when creating shared folder)
    /media/uuid2/storage2
    connect to storage2 via pc and create media folder in storage2 share
    /media/uuid2/storage2/media
    then create all the other media folders in the media fodler (e.g. Movies, TV, Music, Photos, etc...)
    /media/uuid2/storage2/media/Movies2
    /media/uuid2/storage2/media/TV2
    /media/uuid2/storage2/media/Music2
    etc....


    Lastly, when space gets low in main media drive to bring in media from say the movies folders on the Drive 2.
    On Drive 1:
    cd /media/uuid1/media/Movies
    ln -s /media/uuid2/storage2/media/Movies2 Movies2
    (This creates a symlink file named Movies2. To break the symlink all you have to do is delete the Movies2 symlink file. If you do not know how to use symlinks it is a good thing to learn. Plex will recurse into Drive2 via this symlink and read files from the Movies2 folder and you don't need to add another folder in Plex.)


    The 1st drive I do not use the number 1. But as I add drives I start with 2 then 3 then 4, etc...


    PS- You can add symlinks with downloader plugin.
    PPS- All to easy....

  • The feature that I liked best about greyhole was not just the pooling, but the ability to choose how many file copies for which shares. This concept allows for a nice balance between redundancy and space utilization. For example, I don't really care if my Movies and TV shares are lost due to HD failure, but having local redundancy of my Photos, Music, system backups, etc. is a big plus. With greyhole you can have all of that along with pooling.


    It just didn't work for me, is all. So I decided to get similar benefits from a combination of snapraid and mhddfs. I use exclusions to prevent snapraid from creating parity for less important stuff (Movies, TV), and for the stuff I really never want to lose I have the parity disk plus I will be rsyncing to my offsite server.


    Your method is simple, but provides no redundancy. I have seen you say elsewhere that you run multiple local servers and replicate the data. That's not an option for me.

  • Cogitech, I like the number of file copies with Greyhole too. If you have some things you do not need copied and then other that you need a copy or 2 for safety. It is nice feature. I really like Greyhole but it can be confusing, especially for people new to linux.


    I do what I say above and backup every data disk. It is just simpler and faster for me. I know not everyone can have another server and several more disks. Everyone has to decide what is best for them. But it always good to share ideas. :)


    If I did not have a 2nd server for backup I would use SnapRaid for redundancy and Greyhole for pooling. I would probably still make extra copy of important files with Greyhole.


    One big reason I like the method I use is it does not use very many shared folders or many other uses of the data disks beyond Disk 1. If you have a disk failure and the data disks are used by multiple services it can be a pain to replace a failed disk. The setup works well with CrashPlan too. Maybe I elaborate more on this later.

Jetzt mitmachen!

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