[Linux-cluster] Issue with NFS exports and "loopback" NFS mounts as RHCS resources on attempted resource release
Rajagopal Swaminathan
raju.rajsand at gmail.com
Thu Oct 27 18:02:34 UTC 2011
Greetings,
On Thu, Oct 27, 2011 at 7:56 PM, <Ralph.Grothe at itdz-berlin.de> wrote:
> Hi Raj,
>
> yes, I know this SAP on RHCS document and have thouroughly read
> it (I myself supplied the link in my initial posting).
> But as I already wrote NFS isn't addressed in this document.
> Instead it reads on page 32, Section 8.3.3 "NFS Mounted File
> Systems", 2nd paragraph:
>
> "It is possible but not recommended or supported to run an NFS
> server within the cluster. This
> leads to a re-mount scenario in which the cluster node exporting
> the NFS filesystems remounts
> the same NFS exports. In low memory situations, the NFS server
> and client can
> negatively impact system stability."
>
> I know I should show them SAP admins this paragraph, and ask them
> why they are trying to hold me accountable for not working
> features or being deemed inapt to configure a cluster properly
> for requested layouts the clustering software (viz. its agents)
> isn't fit to deliver.
I would understand SAP's uppity attitude if they are approached.
> But this is the usual battle with users and customers, and I
> already was coerced into violating cluster software just to make
> it work as the customer thinks it ought to work a, couple of
> times.
>
I can understand that too, been there done that.
I apologise for I neither have a cluster nor have aSAP implementation
in front of me to give you a more usable answer.
But then, IIRC NFS can operate on UDP (connectionless) as well as TCP
(connection-oriented and default)
But I will keep quite on this thread further. Till I hear from the
experts from Redhat or SAP. I get paid from neither.
--
Regards,
Rajagopal
More information about the Linux-cluster
mailing list