Date: Tue, 4 Sep 2012 02:51:33 +0200 From: =?iso-8859-2?Q?Edward_Tomasz_Napiera=B3a?= <trasz@FreeBSD.org> To: Ivan Voras <ivoras@freebsd.org> Cc: freebsd-fs@freebsd.org Subject: Re: UFS and ACLs Message-ID: <0BA0AE08-FF4D-422E-A00A-D8AA1146B66A@FreeBSD.org> In-Reply-To: <k20isv$k9q$1@ger.gmane.org> References: <k20isv$k9q$1@ger.gmane.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Wiadomo=B6=E6 napisana przez Ivan Voras w dniu 2 wrz 2012, o godz. = 23:25: > Hi, >=20 > Can someone give an estimated / expected answer on these questions: >=20 > * What is the performance impact (if any) for file systems mounted = with > either of the ACL options: acls, nfsv4acls, in the situation where = most > of the files do not have (and do not need) any ACLs? Performance impact for files that don't have an ACL should be = unmeasurable. ACLs are stored in extended attributes, and that has some impact, but we don't store ACLs unless it's actually required. > * Will the kernel automagically add ACLs (other than the regular Unix > DAC bits) to new files on file systems mounted with acls/nfsv4acls? No, unless there are some inheritable ACL entries on the containing directory. > * The "regular" DAC bits have some nice propagation rules, e.g. new > files created by a user belonging to a group which owns the directory > have the GID of this group. Does this work with ACLs? Yes. Mounting the filesystem with ACLs doesn't change anything at all in the behaviour, until you actually set some ACLs. > * Which is the easier option to use/maintain, POSIX or NFSv4 ACLs? Depends on what you need. Conceptually POSIX ACLs are simpler. --=20 If you cut off my head, what would I say? Me and my head, or me and my = body?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?0BA0AE08-FF4D-422E-A00A-D8AA1146B66A>