[Top] [All Lists]

RE: High disk I/O causes unkillable Processes.(Was Linux + XFS + SCS I

To: "Gonyou, Austin" <austin@xxxxxxxxxxxxxxx>
Subject: RE: High disk I/O causes unkillable Processes.(Was Linux + XFS + SCS I = Problems?)
From: Steve Lord <lord@xxxxxxx>
Date: 05 Nov 2001 09:41:28 -0600
Cc: "'linux-xfs@xxxxxxxxxxx'" <linux-xfs@xxxxxxxxxxx>
In-reply-to: <85063BBE668FD411944400D0B744267A888700@AUSMAIL>
References: <85063BBE668FD411944400D0B744267A888700@AUSMAIL>
Sender: owner-linux-xfs@xxxxxxxxxxx
On Mon, 2001-11-05 at 01:34, Gonyou, Austin wrote:
> DMAPI Strikes again.
> I tested my target system and my Dell 1550 with a kernel that didn't have
> DMAPI in it. Voila!
> All tests, both the Java test and my spew.pl scripts have been working like
> a champ!
> I just removed the append="noapic" from my lilo.conf and now I'm retesting.
> I think it will go
> well there also. Thanks to all the people who've helped me through this. 
> I've got one last question though. Is there a way to pass a parameter to a
> kernel to turn off 
> dmapi at boot time so I don't have to compile a new kernel until I'm able to
> deploy it to all of them?
> Maybe something like append="nodmapi" in lilo.conf? Thanks for your time.
> You all have been especially 
> great! I LOVE OSS!
> -- 
> Austin Gonyou
> Systems Architect, CCNA
> Coremetrics, Inc.
> Phone: 512-796-9023
> email: austin@xxxxxxxxxxxxxxx 

Thanks for your persistance on looking at this one Austin, oddly enough
we had a dmapi enabled kernel go into a cpu loop last week - in the
write path, but I did not associate the problem with your configuration.

As a rule I would keep dmapi off, it is really no use to anyone who does
not have a dmapi application to run on the filesystem. All it buys you
in any other case is more code getting exected.



Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@xxxxxxx

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