[Orca-dev] Re: [Orca-users] Tape Throughput

[list2002 at bias.org]

I'm not sure what you are wanting/expecting from my end.

It appears you want me to try commenting:

  // Use the new raw disk code all the time because it has been very
  // well tested and there are no major bug reports in it.
  // #define USE_RAWDISK     1

And then remove all the code I did for tape measuring.  Is this correct?
I would have to also alter orcallator.cfg if I did this though?

It also appears that you have an idea on how to better code the tape
throughput measurement through separate routine.

Thanks,
Liston

On Thu, 14 Nov 2002, Blair Zajac wrote:

> [deleting Orca-users, adding Orca-dev]
>
> Liston Bias wrote:
> >
> > I tried the diff -u option and my diff did not recognize it.  Is there a
> > special diff I should be using?
>
> Liston,
>
> Finally looking at your tape patches for inclusion into orcallator.se.
>
> The diff that supports -u is the GNU diff.
>
> >
> > I have copied all the files for the change to:
> >
> >    http://www.solarisadmin.com/orca/tapes
> >
> > I made the changed to 1.35 cfg/ce files and called them 1.36.  You can,
> > ofcourse, change this to what number you want.  For the purposes of diff,
> > I was trying to:
> >
> > # diff -u orcallator.cfg-1.35.txt orcallator.cfg-1.36.txt > patch_cfg-1.36.txt
> > # diff -u orcallator.se-1.35.txt orcallator.se-1.36.txt > patch_se-1.36.txt
> >
> > I testing these changes this morning without any problems.  I have
> > customized my cfg files here removing a lot of graphs I don't want, so I
> > have tested for extented periods of time.
>
> In your patch, you add
>
>  #ifdef USE_RAWDISK
>    for (i=0; i<RAW_disk_count; ++i) {
> -    // Block the listing of tape drives for now.
> +    // Record tape drive st devices differently than regular disk devices
>      if (RAW_disk[i].short_name[1] == 't' && RAW_disk[i].short_name[0] == 's') {
> +      tape_count++;
> +      total_treads    += RAW_disk[i].reads;
> +      total_twrites   += RAW_disk[i].writes;
> +      total_treadk    += RAW_disk[i].kreads;
> +      total_twritek   += RAW_disk[i].kwrites;
> +      put_output(sprintf("tape_runp_%s", RAW_disk[i].long_name),
> +                 sprintf("%16.5f", RAW_disk[i].run_percent));
>
> This only runs when USE_RAWDISK is defined.  But at the top of your
> file you have this:
>
> // Version 1.36:   Aug 23, 2002 Add measurement for tape (st devices)
> //				that appears to have been removed in Ver 1.28b4.
>
> I did a diff between orcallator.se 1.27 and 1.28b2 and this added
> the RAWDISK code, which was defined by default.
>
> I'm guessing that the rawdisk code turned off the tape statistics.
>
> So could you try commenting out the line
>
> #define USE_RAWDISK		1
>
> in your orcallator.se and see if the tape is still being measured?
>
> I'd like the tape statistics to be recorded regardless of the USE_RAWDISK
> value.  I'm thinking of putting this code into its own function,
> named measure_tape().
>
> >
> > I also modified the verbage (change disk to st tape drives) on the
> > orcallator.html docs page and include some daily graph example for host16.
> >
> >   http://www.solarisadmin.com/orca/tapes/orcallator.txt
> >
> > I did not bother attempting to duplicate all disk stats for tapes as we
> > have no need for them here.  The tape stats that we do have are helpfull
> > for identifying issues with our backup servers.  The sample graphs are
> > from an E420R attach to an L700 backup unit (8 drives).
>
> Cool.  I'll put this on the Orca site.
>
> >
> > I noticed you have no indication in the .cfg what version it is.  Is there
> > a reason you chose not to include this information?
>
> I can put a version into the next checkin.
>
> Best,
> Blair
>
> --
> Blair Zajac <blair at orcaware.com>
> Web and OS performance plots - http://www.orcaware.com/orca/
>

============================================================================
- Liston Bias                       "Anger" is just one letter short
  Alumnus of Oklahoma State Univ    of danger.
  Alumnus of Florida State Univ                       -- Anonymous

  bias at pobox.com
  http://www.pobox.com/~bias
============================================================================