Hi Tim,
The HA-NFS Agent doesn't "break" if this part doesn't work.
It relies on this kstat for dealing with high load conditions and
to avoid "false negatives" on determining if the NFS service is
healthy.
So, while you should be able to limp along with that error
message, for any serious deployment of HA-NFS, one would have to
have that bug in Nevada fixed.
One comment below:
Perkins, Tim wrote:
> I?m in the process of trying to setup a single node cluster with NFS and
> Samba resources. I?ve managed to get Samba up and running great? and I
> have NFS running, but I?m being spammed with errors in my
> /var/adm/messages. The error is ?[ID 176151 daemon.error] Unable to
> lookup nfs:nfs_server from kstat:No such file or directory?.
>
>
>
> At first I thought that maybe it was some sort of incompatibility
> between Cluster Express 9/08 and snv_99 (since all documentation points
> to running CE9/08 with snv_97), so I downloaded and compiled the latest
> ohacds source and installed the latest SUNWscnfs package. That didn?t
> help, so I turned to google.
>
>
>
> I managed to find a forum thread at
> http://forums.sun.com/thread.jspa?messageID=9495551. I also found this
> bug http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6418557.
> Which basically states that instance 0 of nfs in kstat no longer exists.
>
>
>
> I found http://docs.sun.com/app/docs/doc/816-2028/6m8dh0koc?a=view
> which covers an error that is close to mine (at least the user readable
> part? not the error ID). It suggests I run ?kstat -m nfs -i 0 -n
> nfs_server -s calls?. After browsing through google a bit more I ended
> up running into the hanfs_util.c source. I found the line of code that
> does the kstat
> (http://src.opensolaris.org/source/xref/ohac/ohacds/usr/src/cmd/ha-services/nfs/hanfs_util.c
>
> line 129 is `ks
> <http://src.opensolaris.org/source/xref/ohac/ohacds/usr/src/cmd/ha-services/nfs/hanfs_util.c#ks>
>
> = kstat_lookup
> <http://src.opensolaris.org/source/s?defs=kstat_lookup>(kc
> <http://src.opensolaris.org/source/xref/ohac/ohacds/usr/src/cmd/ha-services/nfs/hanfs_util.c#kc>,
>
> "nfs", 0, "nfs_server");?).
>
>
>
> I?m wondering if I change kstat_lookup
> <http://src.opensolaris.org/source/s?defs=kstat_lookup>(kc
> <http://src.opensolaris.org/source/xref/ohac/ohacds/usr/src/cmd/ha-services/nfs/hanfs_util.c#kc>,
>
> "nfs", 0, "nfs_server"); to kstat_lookup
> <http://src.opensolaris.org/source/s?defs=kstat_lookup>(kc
> <http://src.opensolaris.org/source/xref/ohac/ohacds/usr/src/cmd/ha-services/nfs/hanfs_util.c#kc>,
>
> "nfs", 3, "nfs_server"); if I?ll break anything? I?m also wondering if a
That would basically mean you are only counting a subset of
the NFS traffic in determining if your NFS server is healthy. Not
a very appetizing prospect. I personally would just live with the
beast which i understand (that i don't have full coverage for my
NFS server), rather then go along with something which might be
incomplete.
OTOH, this is the beauty of Open Source. YOU take a look at what
monitoring you are missing out by not using the "aggregate" kstat
(instance 0), and YOU determine if YOU are OK with it...
Gotta love OpenSource, very empowering, isn't it? :-)
> fix for this already exists?
Not that i personally am aware of.
Regards,
-ashu
>
>
>
> Thanks,
>
> --Tim
>
>
>
> P.S. Thanks Madhan and Nicholas for your speedy reply regarding my lofi
> globaldevices question! J
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> ha-clusters-discuss mailing list
> ha-clusters-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/ha-clusters-discuss
