we had some weirdness...sometimes mounts not responding in a timely fashion...
I saw three bugs fixed in 8.1.2P1 that may have helped:
http://support.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=638155
http://support.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=651528
http://support.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=653927

Can't say for sure if this is just CMODE or CMODE & 7mode, but since they
share a lot of the code, I suspect it could be both.

Realize, we have a huge amount of tiny files which places an exorbitant load
on our NetApp Heads. We see performance issues before most people.

--tmac

Tim McCarthy
Principal Consultant

          

Clustered ONTAP                                                        Clustered ONTAP
 NCDA ID: XK7R3GEKC1QQ2LVD        RHCE5 805007643429572      NCSIE ID: C14QPHE21FR4YWD4
 Expires: 08 November 2014                 Expires w/release of RHEL7      Expires: 08 November 2014



On Thu, Feb 28, 2013 at 9:21 AM, Alexander Griesser <ag@anexia.at> wrote:
Hey Tim,

can you rlaborate on the NFS issues? Are they only affecting cluster mode or also 7-mode installations?

bye,

Alexander Griesser

System-Administrator

 

ANEXIA Internetdienstleistungs GmbH

 

Telefon: +43-463-208501-320

Telefax: +43-463-208501-500

 

E-Mail: ag@anexia.at

Web: http://www.anexia.at

 

Anschrift Hauptsitz Klagenfurt:Feldkirchnerstraße 140, 9020 Klagenfurt

Geschäftsführer: Alexander Windbichler

Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601


Am 28.02.2013 um 15:17 schrieb "tmac" <tmacmd@gmail.com>:

Hey Steve.

I have three Clustered ONTAP systems. As you probably know, it is very different from 7-Mode,
yet there are many similarities. It *is* useful to have a 7-mode background (although not necessary).

Have you looked through the Online Docs for 8.1.2 Cluster mode? All that is in there in their own sections.
(File & protocol guide)

(SysAdmin Guide)

junction paths are where the individual volumes in ONTAP are placed. 
If I have volumes root (aka /),data, corpdata, exp and web, they might look like this

--->root (/)
       ---> data (/data) ---> corpdata (/data/corpdata)
       ---> exp (/exp)
       ---> web (/web)

From a client, I can mount/map to anywhere in the global namespace


Export-policies replace the /etc/exports file. You can define one or more policies that affect how
volumes are mounted. I can have one policy that lists everything and assign it to all volumes
or I can have many to fit different scenarios. (I use the latter). You typically host-match or some type,
like 192.168.1.0/24 (to allow all hosts on the subnet) or 7.1.4.64/26 (for only the 63 hosts on this net)
or even netgroups from NIS. An Export policy is created and then you put rules in the policy...



Of course, the other big thing is pNFS with Clustered ONTAP. If you have clients that support it
(i.e. RHEL6.2 and higher and some other Fedora kernels) you can turn it on and use it. pNFS
separates the data and the meta-data paths. You can hit any node in the cluster for the meta-data and
behind the scenes redirects you to the origination server for the data (i.e. direct data path) so you get a 
potential performance boost by not using the backend cluster network for data.

hope this helps.

By the way...8.1.2P1 has some very nice NFS fixes in it. we have seen issues vanish since the update
two days ago (we were seeing unexplained issues daily!)

--tmac

Tim McCarthy
Principal Consultant

          

Clustered ONTAP                                                        Clustered ONTAP
 NCDA ID: XK7R3GEKC1QQ2LVD        RHCE5 805007643429572      NCSIE ID: C14QPHE21FR4YWD4
 Expires: 08 November 2014                 Expires w/release of RHEL7      Expires: 08 November 2014



On Thu, Feb 28, 2013 at 8:43 AM, Steve Losen <scl@virginia.edu> wrote:

Hi Folks,

Can anyone point me to some good documentation for the above?
I've looked on the Netapp Support site at the man pages and admin
guides and they cover the command options, but I haven't found
anything explaining how all this stuff works together.  I understand
that the purpose is to provide a unified namespace, but I would
like more information about what is legal/illegal, recommended
practices, etc.

I there a white paper on this?

Steve Losen   scl@virginia.edu    phone: 434-924-0640

University of Virginia               ITC Unix Support


_______________________________________________
Toasters mailing list
Toasters@teaparty.net
http://www.teaparty.net/mailman/listinfo/toasters

_______________________________________________
Toasters mailing list
Toasters@teaparty.net
http://www.teaparty.net/mailman/listinfo/toasters