ACL "reappear" periodically on sharedfolder used by NFS

1. offizieller Beitrag

    Hello board,


    I have the follwing setup:
    - a mergerfs pool
    - a sharedfolder Videos in this pool
    - this sharedfolder is used as a NFS share


    Then my emby server (in another VM) is relying on this NFS share. And my android Emby app is supposed to access in direct to the movie through the NFS share.


    My problem is: the nfs share doesn't work unless I reset the permissions (using the resetperm plugin) to pure POSIX rights.


    But somehow, every now or then (no idea of the frenquency yet - couple of hours maybe?), the ACL reappear and break the NFS share again. I have to reset the permissions and it instantly works again.


    Is there a way to debug this?


    Is there a permanent swith to disable the ACL and prevent them from reappearing once and for good?


    Thanks a lot!


    Edit: it may happen after one scheduled rsync job (local) is run.

    • Offizieller Beitrag

    With the information you provided, this is speculation, I'd look closely at the Rsync job. While the Rsync job may seem to be local, if (for example) a foreign file system is mounted with Remote Mount, the files on the receiving end get their permissions assigned by the local host root account. Where foreign volume copies are concerned, the "preserve permissions" switch doesn't work and it should be noted that an Rsync copy between two VM's is not a local copy.


    In my case, file permissions started at the source as root:users and at the destination they where root:root which messed up what I was attempting to do.


    To get past this problem, I put the following in extra options. (While configured as a local copy, the source is a remote mounted foreign volume and the destination for this job is a local folder.) These settings override the create mask and set the local root and users accounts as the owners of the files copied. It can be customized to your permission preferences.


    Thanks,


    One more thing: the rsync job is actually really local, and the folders are outside of the scope of my NFS shared folder).


    I actually now can confirm:
    - reset permissions ==> verified my use-case is working ok, I can access my NFS share
    - then manually launched the rsync job from the WebUI ==> verified my use-case is KO, cannot access my NFS share


    Now, I reset the permissions on the two sharedfolders used by the rsync job, as well as my NFS sharedfolder. ==> verified access is ok.


    I launched the rsync job, and this time, the NFS shared stayed accessible.


    Nota: I'm absolutely not familiar with ACL. I didn't apply them myself, but I just migrated from my former Ubuntu-based NAS (keeping the disks and permissions as is, though never using ACLs either).


    I think there is still a bug somewhere.

    • Offizieller Beitrag

    With your Rsync job local AND having the flags for preserve permissions and ACL's set - resetting the permissions on the shared folders used by the Rsync job may have fixed the problem. Time will tell. Let us know.


    (BTW: I'll be out of town until next weekend.)

Jetzt mitmachen!

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

Benutzerkonto erstellen Anmelden