xfs
[Top] [All Lists]

Re: How risky is 2.5.x?

To: linux-xfs@xxxxxxxxxxx
Subject: Re: How risky is 2.5.x?
From: Robin Humble <rjh@xxxxxxxxxxxxxxxx>
Date: Mon, 11 Nov 2002 00:30:17 -0500
In-reply-to: <20021110193742.70a3be2b.lloy0076@xxxxxxxxxxx>; from lloy0076@xxxxxxxxxxx on Sun, Nov 10, 2002 at 07:37:42PM +1030
References: <87wunlsvri.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxx> <20021110193742.70a3be2b.lloy0076@xxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.2.5.1i
On Sun, Nov 10, 2002 at 07:37:42PM +1030, David Lloyd wrote:
>It's a development kernel. I lurk on the linux kernel list and there are
>"issues" happening with it all the time and it changes all the time. It's
>a "use only if you are really testing" type thing.

There are 'issues' with all kernels. If there wasn't, then 2.4
development would have stopped at 2.4.0 :-)

I've been running 2.5 XFS since 2.5.43 and have seen no corruption or
crashing. I sometimes get an oops when it's trying to power-off the
machine, but that's well after all filesystems have been unmounted.
It's also possible a non-standard module (mga_vid.o) is to blame.


One bug that I have seen with both 2.4.19-xfs and 2.5.*-xfs is that the
rpm database sometimes gets 'locked' so that no rpm queries can be done
as root. Other users can do an 'rpm -qa' with no problems... not sure if
this is a generic rpm problem an xfs (file locking?) issue.
It happens during rpm -Uvh when the install doesn't complete but
instead hangs forever and needs to be kill -9'd. After that strace
shows that root can't open /var/lib/rpm/Packages and it loops forever
in select() waiting to access it.
Its happened for me with RH7.3, RH8 and with 2.4.19-xfs cvs kernels and
current 2.5 cvs kernels.
Has anyone else seen behaviour like this?

cheers,
robin


<Prev in Thread] Current Thread [Next in Thread>