pcp
[Top] [All Lists]

pcp updates: probing rejig, python pmcc, qa

To: pcp@xxxxxxxxxxx
Subject: pcp updates: probing rejig, python pmcc, qa
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue, 3 Jun 2014 23:57:40 -0400 (EDT)
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1485036797.19840311.1401844672412.JavaMail.zimbra@xxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: fltK0z85cWVDAzuMggFaPJYhSXz8lw==
Thread-topic: pcp updates: probing rejig, python pmcc, qa
Changes committed to git://git.performancecopilot.org/pcp.git dev

 qa/734                       |    5 
 qa/979                       |    4 
 qa/980                       |    2 
 qa/src/test_pmcc.python      |   20 +
 src/libpcp/src/check-statics |    5 
 src/libpcp/src/probe.c       |  461 ++++++++++++++++++++-----------------------
 src/python/pcp/pmcc.py       |   45 ++--
 7 files changed, 268 insertions(+), 274 deletions(-)

commit bdef24031101e2f87c2056cda02d45475b9af22c
Merge: 7446d0d 64a7394
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Wed Jun 4 11:13:16 2014 +1000

    Merge branch 'brolley/dev' of ../pcpfans into brolley-merge

commit 7446d0d46ab2d1c1d610d49b4d05511018d53878
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Wed Jun 4 10:59:03 2014 +1000

    Ensure test qa/739 attempts to run without down-rev remotes
    
    Test 738 exercising context -> hostname issues in pmstat and
    libpcp can still do useful work without a downrev host with
    no pmcd.hostname metric.  Provide a fallback for this case so
    that the test still runs even when no pre-3.8.4 hosts exist.

commit a8a92b832de3ff943a7702ad62903fc032b1d0fd
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Wed Jun 4 10:56:44 2014 +1000

    Improve handling of archives and errors in python pmcc module
    
    Handling of errors in pmcc is much proved - previously it just
    printed a message to stdout for all caught exceptions and then
    continued right on its merry way (!).  This typically resulted
    in promptly getting hit by a bus two or three lines on, due to
    some assumptions about earlier success in setting up some data
    structure.
    
    Particularly problematic was archive replay reaching the end of
    the archive, which is a natural and expected "error" condition.
    Adding to the dilemma is the way we have good context for doing
    error reporting at the initial exception catching site.  Handle
    this by now printing to stderr, then raising SystemExit with an
    exit code indicative of normal(EOL) or abnormal failure.  Then,
    we catch this in the top level run() interface and return that
    code to the script, which can deal with it as it will - usually
    simply by exiting with that code, I expect.
    
    Test qa/980 now passes.  Updated test 979 (live mode) to use a
    shorter sampling interval while in that neck of the woods - to
    complete more quickly but also exercise interval setting.

commit 64a7394458a66cf3d4a4559a7290b421a0994c07
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date:   Tue Jun 3 15:54:21 2014 -0400

    src/libpcp/src/probe.c: Ensure that mutexes are always initialized.
    
    Even if we're only using the main thread for probing.
    Also fixed a few typos.

commit 0642129107c88820fe9924c0320cf3898008287e
Merge: c8e9b3d fa5f6da
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date:   Tue Jun 3 12:03:52 2014 -0400

    Merge remote-tracking branch 'origin/dev' into brolley/dev

commit c8e9b3dc926d0285db06d94207ec36f29f8e3678
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date:   Tue Jun 3 12:01:54 2014 -0400

    Implement fche's threading algorithm for active probing.

commit fa5f6daa4cd9fc4c9d34c71c864a8b572e7c0792
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Jun 3 19:13:29 2014 +1000

    Switch pmcc tests 979 and 980 to using regular pcp options

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates: probing rejig, python pmcc, qa, Nathan Scott <=