Poly 2.5.4 available

From: Alex Rousskov (rousskov@ircache.net)
Date: Mon Jul 24 2000 - 17:30:18 MDT


Hi there,

    Web Polygraph 2.5.4 is available. The change log is attached.
        http://polygraph.ircache.net/sources/

    This release includes primarily bug fixes. It is released today
according with the third cache-off schedule. We probably fixed all
serious PolyMix-3 bugs that we had sufficient information about. There
are about three outstanding problem reports left. At this time it is not
clear whether those problems are Polygraph-related and/or whether the
corresponding bugs are fixed in 2.5.4.

    I would like to believe that version 2.5.4 is stable, but the amount
of last days changes was significant. Keeping that and the outstanding
PRs in mind, I think it would be _prudent_ to wait for 3-4 days before
officially naming the bake-off version. We hope that cache-off
participants will submit enough 2.5.4 reports by Tuesday morning to make
the final call.

    Finally, these few extra days will allow us to implement the IP
allocation change the Duane described on this list. That change is not
visible to run-time Polygraph code, but if you have any objections
please let us know ASAP.

    While waiting for your reports, we will work primarily on
documentation and tools for experiment management and log analysis.

Alex.

    [ bcc-ed to the bakeoff list ]

------------- change log -----------------
2.5.3 -> 2.5.4
20000724
    - added "special_req_factor_beg" and "special_req_factor_end"
      fields to Phase specs to control the presence of special
      requests (IMS, Reload, etc.) in the generated stream (PGL)
    - identify cases when HTTP race conditions lead to connectivity
      errors on used-to-idle persistent connections; retry
      transactions when such errors occur; log the number of retried
      transactions (that otherwise do not affect statistics)
    - eliminated most of the special (non-Basic) requests during the
      fill to shorten the fill phase (PolyMix-3)
    - when a phase meets its goal, wait until all remote phases are
      ready before proceeding
    - added the "--sync_phases <bool>" option to disable auto phase
      synchronization if needed
    - log statistics about oid generation algorithm
    - distinguish "hot" requests in stats samples
    - polished stats samples and made them smaller
    - report WSS in terms of bytes (in addition to object count); be
      careful when interpreting the values -- adjust for embedded
      objects and other hosts in the bench
    - complain if server agent port is not specified
    - upgraded current and required log versions from 2 to 3; log
      versions were not upgraded for a while despite format changes,
      unfortunately
    - warn about log extractor <--> log producer versions mismatch
      when reading a log
    - fixed typo in X-Srv-Sets header parsing code that led to
      growing WSS for public (shared) objects, resulting in
      lower-than-expected HR
    - fixed ./CdbBodyIter.cc:43: assertion failed: '!theInjector'
    - fixed coredumps when all configured phases are ended and
      Polygraph adds a "cold" phase for the last transactions to
      finish and coredumps with segmentation fault
    - made reportgen.cfg dependent on ./configure parameters



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:14 MDT