[5717] in Release_7.7_team
Re: openafs lock fix
daemon@ATHENA.MIT.EDU (William Cattey)
Sun Feb 18 18:48:18 2007
In-Reply-To: <243FDA14-E401-4879-B52F-FA63FFEAA73B@mit.edu>
Mime-Version: 1.0 (Apple Message framework v752.2)
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
Message-Id: <F4C762C5-8EF7-4FC5-970A-75D7C35A56CE@mit.edu>
Cc: Robert A Basch <rbasch@mit.edu>, release-team@mit.edu
Content-Transfer-Encoding: 7bit
From: William Cattey <wdc@MIT.EDU>
Date: Sun, 18 Feb 2007 18:48:11 -0500
To: William Cattey <wdc@mit.edu>
X-Spam-Flag: NO
X-Spam-Score: 0.00
I have read the thread carefully.
Several times.
And now my head hurts, but I think I understand the situation. The
understanding of how to do clien-side byte range locking under Linux
is still evolving. A couple times they thought they had it sorted
out. One of those times it actually created a situation where the
kernel would panic. (This is likely what panic'de the kernel on our
1.4.2 Linux systems.)
They decided to take client side byte range locking off the 1.4.3
feature list.
So they backed out the code that was causing the panic, but also
backed out the code that would do the byte range locking. This means
we can take 1.4.3 when it comes out for Linux with some confidence
that it won't panic where 1.4.2 did. But it also means we should not
expect a fix for locking.
-wdc
----
William Cattey
Linux Platform Coordinator
MIT Information Services & Technology
W92-176, 617-253-0140, wdc@mit.edu
http://web.mit.edu/wdc/www/
On Feb 8, 2007, at 9:23 PM, William Cattey wrote:
> I saw email about a panic problem and remedy in the openafs-devel
> list, but I've not yet read the thread carefully. I believe it's
> relevant, and Jeff Altman, when I mentioned we had a panic
> heisenbug said to try 1.4.3-rc2 when it is released.
>
> I'll update release-team when I've read the thread carefully.
>
> -wdc
>
> ----
>
> William Cattey
> Linux Platform Coordinator
> MIT Information Services & Technology
>
> W92-176, 617-253-0140, wdc@mit.edu
> http://web.mit.edu/wdc/www/
>
>
> On Feb 8, 2007, at 6:15 PM, Robert A Basch wrote:
>
>> An update on an old problem...
>>
>> A fix for the openafs fcntl() lock bug is going into 1.4.3rc2 (I
>> have not tested it). Deploying this is probably not so urgent,
>> as the firefox wrapper tries to deal with the profile remaining
>> locked after the program exits. (It's also unclear whether
>> 1.4.3 will fix the crashes seen with 1.4.2...).
>>
>> Bob
>