pcp
[Top] [All Lists]

pcp updates - mostly QA

To: pcp@xxxxxxxxxxx
Subject: pcp updates - mostly QA
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu, 06 Nov 2014 08:08:05 +1100
Delivered-to: pcp@xxxxxxxxxxx
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.2.0
Changes committed to git://git.performancecopilot.org/kenj/pcp.git dev

 qa/518                |   51 ++++++++++++++++++++++++++++++-----------------
 qa/518.out            |    8 +++----
 qa/553                |    2 -
 qa/763                |    1 
 qa/841                |    1 
 qa/843                |    1 
 qa/985                |    2 -
 src/pcp/pcp.sh        |    1 
 src/pmcd/src/client.c |    4 +++
 src/pmcd/src/dopdus.c |   54 +++++++++++++++++++++++++++++++++++++++++++-------
 10 files changed, 94 insertions(+), 31 deletions(-)

commit 271e99450d614692db688cfb6ec11f11cb5527c5
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 08:06:51 2014 +1100

    pmcd: additional diagnostics in the area of user/group authorisation

commit 4f090f19d352c941c56874a353fc1121ccf6c8e9
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 07:57:17 2014 +1100

    pcp(1): add missing sort for -P output
    
    With multiple pmie's running, pcp -P could fail because the inputs
    to one of the (many) join commands is not not sorted.
    
    Found in the rework of qa/518.

commit b1549077c2e101a73278ffe86249cc5dc6097361
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 07:01:44 2014 +1100

    qa/518: better checking for the expected number of rule evaluations
    
    Change timing to revert to older duration (4sec, not 12sec).
    Avoid use of pmsignal all together, so pmie exit is guaranteed.
    Tolerate small differences in the number of rule evaluations.

commit 9c6480f97d20e5369d347a5535e859becc2d29c3
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 06:58:38 2014 +1100

    qa/832: remove pmns before exit

commit 9ecebec90605bf25af4ea73d31fd4e4a5cfd5df4
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 06:55:50 2014 +1100

    qa/841 - remove domain.h.python before exit

commit 3dde6954f6f042ef5c11d255d6094ea3cc33c9ab
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 06:54:02 2014 +1100

    qa/763 - remove domain.h.perl and pmns.perl before exit

commit 5c13e3fe647822e27b18c862666784e89705b03d
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 06:47:27 2014 +1100

    qa/553 - remove ./gluster.log before exit

commit 1bb003afe463f7aeb118b96c7cf451176ff1e218
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Thu Nov 6 06:45:48 2014 +1100

    qa/985 - remove ./dmcache.log before exit

commit a4393b405e9d3d449d93a40313dfae6d2a2c0d81
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Fri Oct 31 16:03:37 2014 -0400

    qa/518: stretch timing to run more reliably
    
    This test case involves a race between pmie running rules a certain
    number of times, and a separate task waiting (by time) to match the
    results via /usr/bin/pcp.  It was observed as flakey, not always
    winning the race.  This version stretches the pmie time and fine-tunes
    the waiting interval to give the race a larger (multi-second) window
    for success.

commit 31e5a3e1db23bc1755b39612c11fd3aeb18fc96c
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Fri Oct 31 16:02:03 2014 -0400

    qa/518:
    
    It was observed that the $sudo pmie invocation didn't always
    receive the closing signal at this test case, leading it to
    hang.  Now we kill it harder, both by pid and later by name.

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates - mostly QA, Ken McDonell <=