On Thu, 2012-02-16 at 16:15 -0500, Andy Adamson wrote:
> On Thu, Feb 16, 2012 at 3:50 PM, Myklebust, Trond
> <Trond.Myklebust@xxxxxxxxxx> wrote:
> > On Wed, 2012-02-15 at 17:50 -0500, andros@xxxxxxxxxx wrote:
> >> From: Andy Adamson <andros@xxxxxxxxxx>
> >>
> >> Session statistics are needed for performance characterisation and debugging
> >> recovery and migration.
> >>
> >> Only gather forechannel statistics as the backchannel has one session slot.
> >>
> >
> > Could this be done better using tracepoints? I'm not sure that I see how
> > a typical system admin would use this kind of file, so it looks more
> > like a debugging aid.
>
> It is similar to the mountstats xprt stanza with waitq lengths, but I
> see your point. I'll look into a tracepoint version.
>
> Any suggestions of other session info to gather?
* How many times have we reset the session?
* How many slot recalls have we seen?
* How many NFS4ERR_CONN_NOT_BOUND_TO_SESSION?
* How many NFS4ERR_BADSESSION/DEADSESSION?
* How many NFS4ERR_BADSLOT/BAD_HIGH_SLOT?
--
Trond Myklebust
Linux NFS client maintainer
NetApp
Trond.Myklebust@xxxxxxxxxx
www.netapp.com
��.n��������+%������w��{.n�����{��w��� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
