At BETSOL/Zmanda, we're not supporting FreeBSD despite its long and storied past. It should still be fine for the OSS version?
I'm not sure how FreeNAS serves up the disks and how he's mounted them to see permissions fail on the FreeBSD Amanda-client. That's why I didn't reply. If the permissions are failing... is that from a mount of the direct iSCSI disk that's mounted (NTFS I guess?) or is that something else like CIFS and its mounted via Samba?? What FreeBSD kernel is enforcing these 'Windows ACLs' when you see Amanda fail? > On 5/9/19 11:32 AM, Gene Heskett wrote: > Perhaps Chris or one of us can help you. I hope so .. but it's a bit of a shot-in-dark. > It sounds like a permissions problem, and part of amanda's security model > is to not run anything with more permissions than it takes to do that > job. You will need an amanda, amandabackup, or backup user and that user > must be a member of the disk group, or I assume a similar group on BSD. That was my guess. Change (in FreeBSD?) to become amanda or amandabackup and see if the same power that root has is still there. > Amanda must be built and owned by amanda:amandabackup or backup, maybe > even disk for its group. It also needs and .amandabackup file with 0600 > perms as part of its security model. > > There are of course, stuff my old wet ram doesn't recall, and that > someone else can chime in to help. There's a lot of stranger things needed to get a read-everything backup client to work. It is a strange 'janitor-with-all-the-keys' situation. > and I have close to zip experience with autogen. You'll be gratified to know 'autogen' really just a small shell script that (I bet) was hand made a while ago. It jumps over to source from config/set_full_version ... and then uses those files and versions-picked to start things. All it really is is a wrapper for the autotools stuff. It's been around a long long while. The last change made to it (from trunk) was marked as Nov-2014 ... and its relative position is 630 commits before the 3.4.x branches start. This was around the time community-3.3.7 was about to come out > We are no longer using svn, its at git-hub so thats the first thing that > needs fixed. Beyond that, its going to have to be someone with git > perms to fix it. Well, in our defense ... the pull requests were sitting around a long long while. I'd estimate many people were letting it languish and only the elite few were trying to fix anything. It needs to keep going, however. > And if more help isn't forthcoming from Betsol, this > may be The Fork. Dustin Mitchel might be able to help if he has the > time since much of the 4.0.0alpha code came from his work on it years > ago. I couldn't build his last 4 or 5 snapshots, then his grant ran out, > or something. He's still replied to many things recently. Probably not a ton of time. Overall we're counting some of these platforms (FreeBSD) to be more centered in the OSS world ... but, yes, they should hopefully be able to build and run still. > I just printed Chriss's last message, so I am going to follow his to the > letter and see how far I get, the last time I either didn't follow > exactly or I failed. Hopefully the dain-bramage (yes) that comes from using git won't attack too many. When you get past some of that, its amazing speed and power starts to be rather addicting. > Has anyone else succeeded??? Here's to finding out. This is, as mentioned, only for 3.5.1. > Copyright 2019 by Maurice E. Heskett > Cheers, Gene Heskett
