possible Bug in mergerfs?

  • My English isn´t so goot but i hope you cand understand me.
    :)
    I´ve a problem......i use Snapraid / mergerfs .... 4 data-hdd´s á 3tb and two parity - hdd´s á 3tb
    In the last few days we are "fighting" @Technikaffe.de into the forum, that my parity - hdd´s flew over .... they were at 100% full-filling level.
    In the union-filesystem configuration stands "existing path, least free space - 4G free - that was the first entry. In the meantime there stand 20GB free.



    BUT
    xattr -l /media/c05db815-36ba-487f-8613-a8e72b50d579/.mergerfs
    .... shows only 4GB free / minimum space:

    The friends from Technikaffe.de - forum do not why it is - and they thought about a possible bug inside?



    The problem is, that my parity-hdd´s flew over and no sync were possible. I had to move manually (!) files from one data-hdd to an other - and normally Snapraid / mergerfs does that automaticly.



    Could you help me to solve that problem, please?
    :)


    :*

  • What files are on your snapRAID parity drives?

    --
    Google is your friend and Bob's your uncle!


    OMV AMD64 7.x on headless Chenbro NR12000 1U 1x 8m Quad Core E3-1220 3.1GHz 32GB ECC RAM.

  • Sorry, but I'm not sure I understand what the problem you're having is.


    epmfs is path preserving so files are only placed in existing directories. If you don't want to manage directories yourself you have to choose a nonpath preserving policy.

  • @trapexit
    That seems to be not correct, is´nt it? I don´t know.....?!
    If you use Snapraid / mergerfs than the pool (!) is shown as one drive from the sight of the user. If one hdd is full, than Snapraid / mergerfs writes the directory structure to the next hdd.
    For instance......
    In my case the fist used hdd for data was hdd 4 - if the next file is too big for that hdd, than uses the system the next hdd - is the target file for \Daten\Spielfile\Drama -> the system puts THAT structure automaticly to the next used hdd
    That was i could see on my data-hdd´s.... look at http://www.technikaffe.de/foru…nc-m%C3%B6glich/&pageNo=2 please ... thread #31 .....
    At the pictures you can see hdd2 (the secons used data-hdd...) hdd4 was the first used data-hdd.
    - at hdd4 you can see the main structure with all directorys / folders
    - as the drive was full -> the system uses the next hdd (2) and wrote the strucure \Daten\Videos\Spielfilme\xxxxxx automaticly - i did not done that .... that means the system has done that
    .... am i not right? That is what i could see.


    The problem is as discribed - the entry of the graphical configuration interface - 20G minimum space - is not to find in the output of
    xattr -l /media/c05db815-36ba-487f-8613-a8e72b50d579/.mergerfs


    ... only 4G stands inside and in that case the system does not preserve 20G as i would like to do .... it preserves only 4G ..... as shown in thread #1
    That means, that hdd4 is going full to 4GB to the end (not to 20GB to the end i would like to do to prevent runnig full of the hdd)


    Could i explain that right?
    :)


    @gderf
    - snapraid.parity -> at the first parity-hdd
    - snapraid.2-parity -> at the second parity-hdd

    • Offizieller Beitrag

    That seems to be not correct, is´nt it? I don´t know.....?!

    He is the author of mergerfs. I think he knows it well :)

    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!

  • @Michelle_Br I'm having a hard time understanding exactly what's wrong or how you have things setup. My suggestion is to change the policy if you don't want or need path preserving behavior.


    If you think epmfs isn't working correctly please explain exactly the layout of the drives by path, sizes, and behaviors.

  • Hello,
    I was answering the original Thread at technikaffe also,so maybe I can describe the problem a little bit:

    • the problem is not the mergerfs mode, if there is no more space (or less than minfreespace) on the drive, the path will be created on a different drive and the file will be created there. She wants to fill up one drive by another. That seems to be working. So do not look further at the mode "epmfs".
    • she is setting minfreespace in the GUI to 20G instead of the default of 4G. This value is than in the /etc/fstab (as mount option). So this looks fine also. The higher value is just to reserve a little bit more of space on each drive, so that SnapRAID has no problem with parity later on.
    • checking the mergerfs parameters shows up, that after a reboot, the minfreespace is still 4G instead of th 20G in /etc/fstab. The value is printed with xattr -l /media/c05db815-36ba-487f-8613-a8e72b50d579/.mergerfs.
    • due to wrong value in minfreesize (but correctly stored in /etc/fstab), the drive fills more than she wants and she gets problems with SnapRAID later on.
    • setting the minfreespace manually to 20G with xattr works fine and then mergerfs works like expected. But of course this is not working over reboots.
    • for me it looks like, that there may be a problem during mounting of mergerfs in OMV3 (it works perfectly for me in OMV2 and a minfreespace of 50G). Or is there maybe something else during reboot which may reconfigure mergerfs (independently of /etc/fstab)?

    my question:
    is there an easy way to look at the mounting parameters of an active mergerfs file system.
    hope, the problem is now a little clearer.


    @trapexit:
    thanks for your excellent union file system. I like it a lot.

    SW: OMV 4.1.35 (Stone burner)
    HW: MSI-B150 ECO, Pentium G4400, 16GB RAM, 128GB SSD, 4TB HDD, 4*8TB SnapRaid (mergerFS)

  • Hi Guys...
    :)


    My problem ist my not so good English. Sometimes it is very difficult for me to discribe the problem well. That´s why thank you @wtuppa for supporting me.


    For myself - i´m also a Linux-beginner. That means i´m not so experienced in console-instructions. For that reason i put the thread from Technikaffe.de because of the console-outputs.

  • @trapexit
    I´ve wrote in the minimum 20GB free with an console instruction - that´s why you can see it in the fstab
    xattr -w user.mergerfs.minfreespace 20G /media/c05db815-36ba-487f-8613-a8e72b50d579/.mergerfs

  • oh ... yes ..... and THAT is the reason for the probs?
    As you know - my Linux experiences are freshly; i´m not so experienced in it.
    I´ve copied the options from another forum-thread - i do not know from where it was.


    But if you mean, that THIS is the problem; that it means, that every argument after that will be ignored ..... why would the directory structure written on the other data-hdds? The structure was written automaticly; not manually by myself. I do not understand that.....

Jetzt mitmachen!

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