Hi Paul,
----- Original Message -----
> Hi,
>
> I have been working on filling the gap with current PCP NFS coverage by
> designing a NFS Server pmda for review/inclusion. It has the same
> coverage for NFS Server a collectl and nfs-stat taking NFS server
> metrics from the data provided by /proc/net/rpc/nfsd interface.
>
At first glance there would appear to be overlap with metrics from the
Linux kernel PMDA - see the nfs.*, nfs3.*, nfs4.* and rpc.* metrics -
are there some additional metrics added here that don't exist there?
If yes, we should consider adding 'em into the kernel PMDA alongside
the others; if no, the QA you've added here is still highly valuable
- maybe it could be switched over to using the existing metrics?
A python implementation of nfsstat(1) using the PMAPI would also be
a handy addition (gives historical reporting, in particular) if you
have any interest in hacking on monitor tools too? Along a similar
vein, pmcollectl(1) could aquire the -s f (nfs) subsystem option to
report NFS stats.
cheers.
--
Nathan
|