<4> post_create: setxatter failed

[Dan Hollis]

On Tue, 31 May 2005, Dave Jones wrote:
>  > Not sure what the point is recommending people test unsupported stuff in 
>  > Fedora if they're going to be closed WONTFIX anyway... does Fedora want 
>  > unsupported stuff tested or not? It would be nice to get an official 
>  > statement on the issue.
> If you hit bugs in JFS/XFS/Reiserfs, you're pretty much guaranteed not
> to get Red Hat folks jumping on those bugs.  There's just not enough
> manpower to attack everything (especially with some of those filesystems
> being very complicated internally), and hence we narrow the scope of bugs
> by limiting the filesystems we class as 'supported'[1].
> The best way forward if you must use those filesystems for whatever reason and
> you find bugs, is to file them upstream. As Fedora aggressively tracks the
> upstream kernel, the faster stuff is fixed there, the faster it gets fixed in
> Fedora.
> [1] Supported is something of a misnomer here given Fedora by its nature
> is unsupported, but for sake of argument think "We'll investigate this
> bug if it gets filed in bugzilla.redhat.com"

Actually the problem here is that these bugs appear to be regressions in 
the fedora _installer_ (which is why they were filed against anaconda). 
That is, they are not related to kernel bugs (you cannot even install 
reiserfs with selinux *disabled*). But they are summarily closed WONTFIX 
which I find troubling.

Something is busted in the installer, a regression from FC3. Silent 
failures, install corruption, etc. It might indicate more serious general
problems lurking about in the installer. But it gets closed WONTFIX 
without even a second thought.

Worse yet I ran into an x86_64 bug yesterday where the kernel would panic 
on the installer if you had unclean xfs partitions in your system (even 
if you were only installing ext3). Why bother reporting when these things 
get instantly closed WONTFIX because they have "xfs" in them?

-Dan