I have an ESTALE problem. That probably means it's been discussed to death,
but after browsing the archive I'm not much the wiser, so...
Here's the scenario:
* RH 7.2 running a 2.4.18 kernel
* Linux boxes acting as NFS clients to a SNAP filer.
* Mount options rw,noac,rsize=4096,wsize=4096,hard,intr,bg
We had a power outage on the filer; the clients stayed up. The clients
returned stale errors, even once the filer had come back up, until we
unmounted and remounted all the mount points.
Is this behaviour as expected, a known problem, an unknown problem, or too
sloppily described to be able to tell?
Regards,
Edward Hibbert
Internet Applications Group
Data Connection Ltd
Tel: +44 131 662 1212 Fax: +44 131 662 1345
Email: [email protected] Web: http://www.dataconnection.com
-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
hi edward-
> I have an ESTALE problem. That probably means it's been=20
> discussed to death,
> but after browsing the archive I'm not much the wiser, so...
>=20
> Here's the scenario:
> * RH 7.2 running a 2.4.18 kernel
> * Linux boxes acting as NFS clients to a SNAP filer.
> * Mount options rw,noac,rsize=3D4096,wsize=3D4096,hard,intr,bg
>=20
> We had a power outage on the filer; the clients stayed up. =20
> The clients
> returned stale errors, even once the filer had come back up, until we
> unmounted and remounted all the mount points.
>=20
> Is this behaviour as expected, a known problem, an unknown=20
> problem, or too
> sloppily described to be able to tell?
we've had several credible reports of this kind of behavior
against many releases of 2.4 recently, so it is a known problem.
but we haven't had a good chance to drill into to it, so we
don't know what is happening.
-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
Is a 'SNAP filer' a NAS box from Snap Appliance or a NetApp filer?
James Pearson
Edward Hibbert wrote:
>
> I have an ESTALE problem. That probably means it's been discussed to death,
> but after browsing the archive I'm not much the wiser, so...
>
> Here's the scenario:
> * RH 7.2 running a 2.4.18 kernel
> * Linux boxes acting as NFS clients to a SNAP filer.
> * Mount options rw,noac,rsize=4096,wsize=4096,hard,intr,bg
>
> We had a power outage on the filer; the clients stayed up. The clients
> returned stale errors, even once the filer had come back up, until we
> unmounted and remounted all the mount points.
>
> Is this behaviour as expected, a known problem, an unknown problem, or too
> sloppily described to be able to tell?
>
> Regards,
>
> Edward Hibbert
> Internet Applications Group
> Data Connection Ltd
> Tel: +44 131 662 1212 Fax: +44 131 662 1345
> Email: [email protected] Web: http://www.dataconnection.com
>
> -------------------------------------------------------
> This SF.net email is sponsored by: ObjectStore.
> If flattening out C++ or Java code to make your application fit in a
> relational database is painful, don't do it! Check out ObjectStore.
> Now part of Progress Software. http://www.objectstore.net/sourceforge
> _______________________________________________
> NFS maillist - [email protected]
> https://lists.sourceforge.net/lists/listinfo/nfs
-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
Sorry, I should have been clearer. Snap Appliance.
Regards,
Edward Hibbert
Internet Applications Group
Data Connection Ltd
Tel: +44 131 662 1212 Fax: +44 131 662 1345
Email: [email protected] Web: http://www.dataconnection.com
-----Original Message-----
From: James Pearson [mailto:[email protected]]
Sent: Wednesday, May 28, 2003 4:16 PM
To: Edward Hibbert
Cc: '[email protected]'
Subject: Re: [NFS] Linux client, SNAP filer, ESTALE
Is a 'SNAP filer' a NAS box from Snap Appliance or a NetApp filer?
James Pearson
Edward Hibbert wrote:
>
> I have an ESTALE problem. That probably means it's been discussed to
death,
> but after browsing the archive I'm not much the wiser, so...
>
> Here's the scenario:
> * RH 7.2 running a 2.4.18 kernel
> * Linux boxes acting as NFS clients to a SNAP filer.
> * Mount options rw,noac,rsize=4096,wsize=4096,hard,intr,bg
>
> We had a power outage on the filer; the clients stayed up. The clients
> returned stale errors, even once the filer had come back up, until we
> unmounted and remounted all the mount points.
>
> Is this behaviour as expected, a known problem, an unknown problem, or too
> sloppily described to be able to tell?
>
> Regards,
>
> Edward Hibbert
> Internet Applications Group
> Data Connection Ltd
> Tel: +44 131 662 1212 Fax: +44 131 662 1345
> Email: [email protected] Web: http://www.dataconnection.com
>
> -------------------------------------------------------
> This SF.net email is sponsored by: ObjectStore.
> If flattening out C++ or Java code to make your application fit in a
> relational database is painful, don't do it! Check out ObjectStore.
> Now part of Progress Software. http://www.objectstore.net/sourceforge
> _______________________________________________
> NFS maillist - [email protected]
> https://lists.sourceforge.net/lists/listinfo/nfs
-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
I believe SNAP servers run Linux with an XFS file system.
If I've got this correct, then with a power out on a Linux Box running
XFS, you can get the the case where file contents are replaced by
'NULLs' - see:
http://oss.sgi.com/projects/xfs/faq.html#nulls
If this happens to the server's /var/lib/nfs/rmtab file, then the server
has lost all records of client mounts after the reboot - hence you get
stale NFS file handles on the clients ...
I believe the cases where NULLs appear has improved since the FAQ note
above (and even more recently in the CVS XFS code) - however, you will
probably have to contact Snap Appliance about this ...
James Pearson
Edward Hibbert wrote:
>
> Sorry, I should have been clearer. Snap Appliance.
>
> Regards,
>
> Edward Hibbert
> Internet Applications Group
> Data Connection Ltd
> Tel: +44 131 662 1212 Fax: +44 131 662 1345
> Email: [email protected] Web: http://www.dataconnection.com
>
> -----Original Message-----
> From: James Pearson [mailto:[email protected]]
> Sent: Wednesday, May 28, 2003 4:16 PM
> To: Edward Hibbert
> Cc: '[email protected]'
> Subject: Re: [NFS] Linux client, SNAP filer, ESTALE
>
> Is a 'SNAP filer' a NAS box from Snap Appliance or a NetApp filer?
>
> James Pearson
>
> Edward Hibbert wrote:
> >
> > I have an ESTALE problem. That probably means it's been discussed to
> death,
> > but after browsing the archive I'm not much the wiser, so...
> >
> > Here's the scenario:
> > * RH 7.2 running a 2.4.18 kernel
> > * Linux boxes acting as NFS clients to a SNAP filer.
> > * Mount options rw,noac,rsize=4096,wsize=4096,hard,intr,bg
> >
> > We had a power outage on the filer; the clients stayed up. The clients
> > returned stale errors, even once the filer had come back up, until we
> > unmounted and remounted all the mount points.
> >
> > Is this behaviour as expected, a known problem, an unknown problem, or too
> > sloppily described to be able to tell?
> >
> > Regards,
> >
> > Edward Hibbert
> > Internet Applications Group
> > Data Connection Ltd
> > Tel: +44 131 662 1212 Fax: +44 131 662 1345
> > Email: [email protected] Web: http://www.dataconnection.com
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: ObjectStore.
> > If flattening out C++ or Java code to make your application fit in a
> > relational database is painful, don't do it! Check out ObjectStore.
> > Now part of Progress Software. http://www.objectstore.net/sourceforge
> > _______________________________________________
> > NFS maillist - [email protected]
> > https://lists.sourceforge.net/lists/listinfo/nfs
-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs