MergerFS+ZFS Questions

1. offizieller Beitrag

    Hi,


    actually I got a mergerfs-pool made out of some 3.5" drives. In futurue I'll go with 2.5" drives and ZFS (without ECC; I know...) Is it possible to insert the ZFS-Pool into mergerfs as well? Or should I stay with mergerfs+snapraid?
    Since it got 10GBe now, I want to use my OMV-Fileserver as my mainstorage for _everything_. I want to store software and games from my client on it. Backups will be made offsite and the important stuff on a second server as well.
    I just want to minimalize the bottleneck. In future even video-editing should be performed on the Fileserver-Disks.
    ZFS has the nice option to add some SSDs as a Cache; sadly mergerfs doesn't support this.

    • Offizieller Beitrag

    If you are using zfs, why use mergerfs at all?

    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!

    Gefällt mir 1
    Zitat

    ZFS has the nice option to add some SSDs as a Cache; sadly mergerfs doesn't support this.


    Because mergerfs is for merging drives for largely WORM data. If you want to create a cached drive and export it to mergerfs then go for it but mergerfs isn't intended to be a caching layer. Most usecases I've heard probably wouldn't benefit from a cache anyway.


    Unless you're attempting to saturate your connection end to end with a single transfer a cache is unnecessary. If you want to send large files quickly your cache will constantly be being pushed to the slow drives which would slow you down. If you're transferring small files why bother with the extra complexity to save a few seconds transfer?


    Also there are ways to use mergerfs and get a cache drive and still be in the spirit of mergerfs' JBOD. Create two mergerfs pools. One with SSDs and one without. Pick a create policy to favor the SSDs (ff, or bind mount the drives as read only somewhere so mergerfs ignores them for writes though that means if you run out of space on your SSDs you'll get out of space errors). Regularly rsync into the SSDless pool and restrict the rsync to files with mtime greater than some timeout.

Jetzt mitmachen!

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

Benutzerkonto erstellen Anmelden