Hi,
I've fixed my problem by commenting out ll_file_noflock from
ll_file_operations on client side,
I think it completely restored the b5 behavior,
as i dont need cluster-wide locks only node wide it should work just fine
for now.
As for a test case....
Appliaction causing the bug was courier-imap daemon running inside my mail
handling VServer guest,
evry time I killed courier-imap daemon it hanged and in the same moment mdt
on the server was dead,
server is in a semi-production state so i would really like it to be
running.
Konrad Gutkowski
----- Original Message -----
From: "Oleg Drokin" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <[email protected]>
Sent: Monday, March 26, 2007 2:09 PM
Subject: Re: [Lustre-discuss] lustre errors
Hello!
On Thu, Mar 22, 2007 at 12:14:20AM +0100, [EMAIL PROTECTED] wrote:
I have recently upgraded my lustre server and clients from 1.6 b5 to b7,
since then im getting those in dmesg:
flock and posix locks are still somewhat flawed, but we are working on it
currently.
You might find a patch from bug 5145 helpful in your situation, or might
be not.
In any case if you can somehow reduce the scope of the problem to some
simple
testcases, that would be really great.
Thanks.
Bye,
Oleg
_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
