[Ocfs2-users] Catatonic nodes under SLES10

[Alexei_Roudnev]

If I know, that FS was not active for the last (for example) 30 seconds, and
all buffers are written back more than 30 secoinds ago, I DO KNOW that there
is not any IO in the pipelines.

At least it can be configurable. 30 seconds - ok, may be default is 2
minutes - some timeout is more then enough. Each scsi system have it's own
time, after which you can say for sure that iut have not any IO in the
pipeline. On the other hand, system reboot dont prevent ALL possible IO
pending as well.




----- Original Message ----- 
From: "Joel Becker" <Joel.Becker at oracle.com>
To: "Eckenfels. Bernd" <B.Eckenfels at seeburger.de>
Cc: <ocfs2-users at oss.oracle.com>
Sent: Tuesday, April 10, 2007 10:19 AM
Subject: Re: [Ocfs2-users] Catatonic nodes under SLES10


> On Tue, Apr 10, 2007 at 07:06:02PM +0200, Eckenfels. Bernd wrote:
> > > It's not at all about what your past activitiy was like. We fence to
> > prevent future activity.
> >
> > There is a lot you can do instead, including SCSI plugging the device o
> > just setting a RO flag in the filesystem (remount-ro-on-error style).
> > There might be times when it is needed to shoot the node in the head,
> > but those are not handled by the self-panic anyway...
>
> By the time you determine you need a node to fence, you do not
> know what I/O it has in its pipeline.  Any I/O that is below the request
> queue can't reliably be stopped in Linux.  If that I/O goes out after
> other nodes have decided the node is gone, it is corruption.  This is
> why fencing has to be absolute.
>
> Joel
>
> -- 
>
> "War doesn't determine who's right; war determines who's left."
>
> Joel Becker
> Principal Software Developer
> Oracle
> E-mail: joel.becker at oracle.com
> Phone: (650) 506-8127
>
> _______________________________________________
> Ocfs2-users mailing list
> Ocfs2-users at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-users
>