[Orca-users] Re: Too many Disk graphs

[Blair Zajac]

JV wrote:

> I know it's a known bug - if you ever move disks or controllers or i/o
> boards I think you're stuck with the duplicates for life. What I would like
> to know from fellow orca users:

JV,

Not any more.  Check out the snapshot in

http://www.orcaware.com/orca/pub/snapshots/

> 1. Has anybody played with sed'ing the interior of a datafile to reduce this
> duplicate graph problem? 
> 
> 
> 2. ever experiment with turning OFF that disk collection parameter? Would I
> comment out something in orcallator.se? Do I have to find and disable it's
> twin in orcallator.cfg or I'll get a bunch of ugly errors during the orca
> cycle run?

You can comment out the

#define WATCH_DISK              1

line in orcallator.se to stop measuring this data completely.

If you want the data but not the plots, comment out the sections in orcallator.cfg.

> How much overhead is there for some of the obscure parameters that perhaps
> your shop does not find very interesting? I am thinking of "skinnying down"
> what orcallator.se collects to reduce the "orca -o orcallator.cfg" cycle run
> times.

Not much overhead at all, except for space.

Check out this section of orcallator.se:

// If WATCH_OS is defined, then measure every part of the operating
// system.
#ifdef WATCH_OS
#define WATCH_CPU               1
#define WATCH_MUTEX             1
#define WATCH_NET               1
#define WATCH_TCP               1
#define WATCH_NFS_CLIENT        1
#define WATCH_NFS_SERVER        1
#define WATCH_MOUNTS            1
#define WATCH_DISK              1
#define WATCH_DNLC              1
#define WATCH_INODE             1
#define WATCH_RAM               1
#define WATCH_PAGES             1
#endif

WATCH_OS turns all these on.  You can either comment out these lines or better 
yet, edit start_orcallator and replace -DWATCH_OS with the -DWATCH_XXXX that you 
want.

Best,
Blair

-- 
Blair Zajac <blair at orcaware.com>
Plots of your system's performance - http://www.orcaware.com/orca/