Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 31 Aug 2018 17:13:33 +0000
From:      bugzilla-noreply@freebsd.org
To:        fs@FreeBSD.org
Subject:   [Bug 212323] tests/sys/acl/01:main fails due to changes in NFSv4 ACL behavior on ^/head
Message-ID:  <bug-212323-3630-xQ28O5uJ1I@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-212323-3630@https.bugs.freebsd.org/bugzilla/>
References:  <bug-212323-3630@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D212323

--- Comment #9 from Harald Schmalzbauer <bugzilla.freebsd@omnilan.de> ---
(In reply to Edward Tomasz Napierala from comment #8)

Unfortunately samba is an extra bunch of issues.
The problems which made me stop using samba for distinct jobs in windows
environments were not only/directly related to ACLs.
I know that some of them have been fixed meanwhile, but I know of others wh=
ich
are new and unresolved (at least on FreeBSD), so even if you end up with a
satisfying NFSv4 ACL overhaul/FreeBSD-standardization, this was only a small
part making samba usable in production environments =E2=80=93 which is not =
the goal of
this problem report (still have to analyze a trace about WindowsServerBackup
failing, compared to windows native implementation of SMB2_02, SMB2_10 and
another version, but couldn't find time and have to start over again at next
attempt =E2=80=93 which probably will never happen because a few hours were=
n't enough
for me to isolate the crucial area last time).

Since it's about a design job rather than fixing, it's best to look at exis=
ting
conventions for other type of ACLs on other operating systems =E2=80=93 Lin=
ux and XFS
comes to my mind and of course NTFS on Windows.  JFS on AIX was more POSIXA=
CL
like, missing the majority of the NFSv4/NTFS flags.
Stearing the unclear semantics and convention into a NTFS friendly direction
would be desireable.
But I never read any posixacl standard also :-(
That said, I also missed reading any (official) NTFSv5 documents and I don't
even know how the inheritance is done in windows. The NTFS versions I start=
ed
with (NT4) didn't provide inheritance and since then I did observations onl=
y.

I'd need to setup a complete artificial test environment with at least 3
windows versions, since I currently have absolutely no idea which combinati=
on
does anything else wrong but ACL related stuff...
Let me think about it over the weekend.  I planed to abuse one cold-standby
system for extended iflib, if_vlan(4) and ctld(8) tests before 12-release,
since there are untracked issues as well (btw. the ctld(8) on 11.2 and
Server2016-initiators turns out to be jumbo frames related).
Let me start materializing this test environment and if things run smooth, =
I'll
extend the FreeBSD machine to serve a r299448 bhyve VM.  For NFSv4 ACLs we'=
re
not bound to real hardware as far as I can imagine at the moment (while ifl=
ib
needs...)
The ESXi Hypervisor is yet to setup too, and undusting the equipment I have=
 in
mind will most likely discover new problems, and this will bring me into big
time troubles.=20
Sorry for my hesitation, but I don't want to promise anything I can't do du=
e to
real job interference...  But I'll try hard and come back if I made first
progress.

Thanks,

-harry


P.S.: samba gained a vfs_freebsd module, which is an attempt to fix various
problems which never affected me as far as I understood.  Haven't found any
documentation about this module, so I need to look into that code to make s=
ure
there's no adaptor to sysutils/libsunacl disturbing test cases. For this
reading my C skills should suffice.

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-212323-3630-xQ28O5uJ1I>