Polygraph 2.6.4 available

From: Alex Rousskov (rousskov@measurement-factory.com)
Date: Mon Jul 09 2001 - 15:24:55 MDT


Hi there,

    Web Polygraph 2.6.4 is available. The change log is attached.
        http://www.web-polygraph.org/downloads/

    This is a minor release that fixes a few 2.6.3 bugs and polishes
old PolyMix-3 workload files. The number of client-side machines was
computed incorrectly for PolyMix-3 in 2.6. The bug led to smaller
working set sizes and other bad side effects.

    Another important change to note is that Polygraph now installs
workload files in /usr/local/polygraph/workloads (default) when you
run "make install". In case of a conflict, Polygraph will overwrite
any files already in the workloads directory. While we have
discouraged users to modify original files, now it becomes dangerous
to do so. Please always copy/rename workload files before modifying
them.

    Finally, folks that wanted to test multiple proxies w/o L4/7
redirection can now do so with the newly added "Robot.proxies" field.
 http://www.web-polygraph.org/docs/reference/pgl/types.html#type:docs/reference/pgl/types/Robot

Enjoy,

Alex.

------------- change log -----------------
2.6.3 -> 2.6.4
20010709
    - added "proxies" field to robot PGL configuration; proxies
      must not be used together with the --proxy command line
      option as they are mutually exclusive; a robot selects a
      random proxy at the configuration time and uses that proxy
      for the entire duration of the test (sticky proxy
      assignment); proxy addresses are distributed evenly (if
      possible) among all robots in the test; individual groups of
      robots (e.g., all robots on one host) may not get an even
      distribution.
    - support "make install" in workload files; "make install"
      will overwrite existing files in the destination directory
      (/usr/local/polygraph/workloads by default); users are not
      supposed to modify original workload files and should
      create copies when modifications are needed
    - made aka robust when no primary IP is present on the interface
    - lognormal mean and standard deviation were not reported
      correctly (Johnson Lee <jl03@gte.com>)
    - synced comments in polymix-3.pg with Bench type changes
    - clientHostCount in PolyMix-3 on 2.6 branch was not computed
      correctly, leading to ridiculously small working set sizes
      and shorter fill phases



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:20 MDT