[Orca-users] Orca Linux

[Sean O'Neill]

At 10:38 PM 3/13/2003 -0600, Liston Bias wrote:
>On Thu, 13 Mar 2003, Blair Zajac wrote:
> > Sean O'Neill wrote:
> > >
> > > At 02:25 PM 3/13/2003 -0600, Liston Bias wrote:
> > > >I just implement procallator data gather for linux and associated
> > > >orcallator.cfg file.  It seems to work great, however, I'm hearing about
> > > >folks "preferring" some of the linux orca graphs vs. solaris.  Do other
> > > >hear this also?
> > > >
> > > >I haven't really looked at the difference yet.  I figure most of the 
> data
> > > >is being collected already (like the flatline CPU count on load 
> graph) and
> > > >I just need to adjust to *.cfg configuration.
> > >
> > > Could you be a bit more specific about what folks apparently like better
> > > about the Linux graphs ?  I'm curious about this.
> >
> > Me too.
> >
> > One of the things I just noticed was that the procollator ethernet
> > bandwidth plots have the loopback interface (lo) which has much more
> > traffic than my two Ethernet cards, so I'm considering removing it
> > from the plot and potentially creating another plot for it.
> >
> > Comments?
>
>In briefly looking, it looks like it has this additional data:
>
>  flat line on load graph

Not sure if this has been added yet.  I think Blair was doing something 
with this.

>  page in/out rates

This is in the current unstable version in the Orca subversion repos.  I 
added this not too long ago.  Not sure when Blair is going to release this 
version.  I added the following:

//                       pgrec/s - page reclaims (include pageout)
//                       pgfrec/s - page reclaims from free list
//                       pgin/s - pageins
//                       pages_in/s - pages paged in
//                       pgout/s - pageouts
//                       pages_out/s - pages paged out
//                       dfree/s - pages freed by daemon or auto
//                       min_fault/s - minor page faults
//                                     (pvm.hat_fault + pvm.as_fault)
//                       maj_fault/s - major page faults
//                       prot_fault/s - protection faults
//                       cow_fault/s - copy-on-write faults
//                       zfod/s - zero-fill-on-demand faults
//                       interrupts/s - interrupts including clock
//                       intrthreads/s - interrupts as threads
//                                       (below clock)
//                       system_calls/s
//                       context_switches/s
//                       invol_switches/s - involuntary context switches
//                       traps/s
//                       forks/s
//                       vforks/s
//                       execs/s
//                       namei/s - pathname lookups
//                       ufsiget/s - ufs_iget() calls
//                       ufsdirblk/s - directory blocks read

You can see graphs of this info at http://chivas.oneill.dhs.org/orca

>  swap in/out rates

I'll look at adding this.

>  Context switches & Interrupts rate

This is in the current unstable version in the Orca subversion repos.  I 
added this not too long ago.   See above.

>  IP statistics (forwarding rate, Inreceive rate, etc)

Take a look at adding this.

>  TCP stats (Active opens, Passive opens, CurrEst rate, etc)

The Solaris version currently tracks the following for TCP:

TCP Bits/sec
TCP Segments/sec
TCP Retrans %/ Duplicates Received %
TCP New Connection Rate (Passive vs. Active)
TCP Open Connections
TCP Resets/sec
TCP Attempts Fail/sec
TCP Listen Drop / sec (Listen Drops vs. Q0 Drops vs. Half Open Drops)

A couple of the links for the sample Linux graphs didn't work so I couldn't 
compare the differences of what may be missing between the two.

>  ICMP stats

Its collecting nothing here that I know of.  IBID on the graphs not working 
for the sample Linux graphs.

>  UDP stats (Indatagrams, noports, inerrors, outdatagrams)

Collecting nothing here that I know of.


--
........................................................
......... ..- -. .. -..- .-. ..- .-.. . ... ............
.-- .. -. -... .-.. --- .-- ... -.. .-. --- --- .-.. ...

Sean O'Neill