Hello Digital UNIX managers,
There is (finally) a new version of 'collect' (and 'collgui') available
on:
ftp.digital.com:/pub/DEC/collect/ (USA)
ftp.digital.de:/pub/DEC/collect/ (Germany)
If you were frustrated with the buggy behaviour of 'collgui' in previous
versions, I recommend getting this one.
The files are:
collect_1.07.tar.gz - the sources, gnu-zipped and tarred (76KB)
COLLECT107SETLD.tar - a setld-installable binary version (2.3MB)
as usual, you will also need perl5 if you want to use 'collgui', the
graphical front end for data interpretation. perl5 is also available
in the same directory as a setld-installable kit called "PERL5003SETLD.tar".
It is 5.6MB large.
If you are curious about 'collgui', you can get some screenshots of
it, and of the gnuplot graphics it produces. They are:
collgui-screenshot.jpg - the control application in perl5/tk with
data selected that corresponds to the
following 2 screenshots
gnuplot-screenshot.jpg - the gnuplot graph of the data
gnuplot-annotated.jpg - the same as above, but with annotated curves
I'm not aware of any bugs in 'collgui'. Please keep in mind, however, that
it is not meant to be competition for the Polycenter Performace Whatever. The
intent was to provide _some sort_ of a graphical tool for analysis of
collected data. The interface may not be particularly intuitive. For those
with special needs, there is always 'cfilt' (the "back end" of 'collgui')
which does the actual work of extracting data from the text output of
'collect'. It can be used to extract just about any data that one might
be interested in. There are now manpages for 'collect', 'collgui', and
'cfilt'.
The following is a list of changes since the last version:
December, 1996 (1.06 - 1.07):
o the binary format has changed! 1.07 will not be able to read
pre 1.07 data files. If this affects you, rename the 1.06
executable to collect1.06 (or something), and move it out
of /usr/opt/COL106/bin if you used the setld kit.
o fixed bug in collgui when two or more objects are selected, and
normalization is enabled. Normalization wasn't being done.
Example: lsm:name=vol01,vol02:rkb/s+wkb/s% didn't work properly.
o fixed bug in collect which affected the selection of disks
with non-null LUNs (mostly on HSZs), such as rza33, rzb33, etc.
o new man pages for cfilt and collgui (mostly the corresponding
READMEs regurgitated, but, like, as manpages, eh?)
o collect now flushes stdout when writing to it, and cfilt will
immediately process data IF no normalization is being done, so
it is now possible to do something like:
collect -i1 -F -sp | cfilt proc+:user=urban:rss
to sum RSS for user 'urban' for all processes owned by 'urban',
and get the results immediately.
o fixed bug in load average which caused all load averages over
2.55 to be misrepresented (collect).
o superficially tested RAID (SWXCR) support. Seems to work!
o in 'collgui', changed the 'Memory Used' expression in the memory
subsystem to use the 'cfilt' expression 'Active+Inactive+Wired'
because 'Active' already contains the UBC pages.
As usual, if you find collect useful, I'd appreciate a note.
Rob Urban
urban_at_rto.dec.com
Received on Wed Jan 15 1997 - 22:37:54 NZDT