pcp
[Top] [All Lists]

pcp updates - github #23, qa and pmlogger review changes

To: pcp@xxxxxxxxxxx
Subject: pcp updates - github #23, qa and pmlogger review changes
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Fri, 05 Jun 2015 17:15:36 +1000
Delivered-to: pcp@xxxxxxxxxxx
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.7.0
Changes committed to git://git.pcp.io/kenj/pcp master

Ken McDonell (8):
      qa/260: improve stability
      pmnewlog: clean up usage message
      pmlogger: revert small part of local context change
      qa/948: add -K exerciser and some error cases
      pmlogger: add comment explaining -o => opts.Lflag usage
      pmlogger.c: initialize linger
      qa/948: extend coverage
      configure: add /bin/sh as possible pid==1 candidate

 configure                   |    3 +
 configure.ac                |    3 +
 qa/260                      |   21 +++++++---
 qa/260.out                  |   16 ++++----
 qa/948                      |   88 +++++++++++++++++++++++++++++++++++++++-----
 qa/948.out                  |   23 +++++++++++
 src/pmlogger/pmnewlog.sh    |   48 +++++++++++++-----------
 src/pmlogger/src/fetch.c    |    2 -
 src/pmlogger/src/pmlogger.c |    8 +++-
 9 files changed, 164 insertions(+), 48 deletions(-)

Details ...

commit 9ab615d4ca7e85501b72587194fbe46c72143aab
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 17:12:59 2015 +1000

    configure: add /bin/sh as possible pid==1 candidate
    
    For https://github.com/performancecopilot/pcp/issues/23

commit 79f88b7107d2c397ff2215b052bb671add4fe81b
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 17:12:29 2015 +1000

    qa/948: extend coverage

commit 7718db828ee37f4d0592118cf385ff06b33ad459
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 16:40:11 2015 +1000

    pmlogger.c: initialize linger

commit 59a03622722bb9c1f6e013f401663892f27fe096
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 11:11:37 2015 +1000

    pmlogger: add comment explaining -o => opts.Lflag usage

commit 104fcd1393569fc52291c3effdc245e7022ea894
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 09:58:49 2015 +1000

    qa/948: add -K exerciser and some error cases

commit 47c15dc36ee7bc29dd2cc7db3e9d16d15f0b5ff1
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 09:55:45 2015 +1000

    pmlogger: revert small part of local context change
    
    Re-instate first argument to __pmEncodeResult() in myFetch().

commit abacb4863ba5c3a2976ae37b7f9ed2c9e73b36dc
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 09:52:09 2015 +1000

    pmnewlog: clean up usage message
    
    - no = for short form options
    - whitespace consistency
    - short and long form for all relevant options
    - add some missing option arguments
    
    e.g. "-n FILE, --namespace=FILE" instead of "--namespace"

commit 124009a0ee2b242dc8f0fadfad69e7c0cc2778f0
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Fri Jun 5 06:28:57 2015 +1000

    qa/260: improve stability
    
    On slow platforms (especially VMs on low powered hosts) the pmie
    startup time may be non-zero, especially where complex rules are
    involved and derived metrics are in use.  This makes the time of
    the first fetch variable and since the metrics are counters the
    values from the firect fetch are still in use at the reporting after
    the second fetch ... this makes the first results unknown (which
    is deterministic) but the second results are not stable in the
    presence of the rate() function because the values and timestamps
    for the firtst fetch are a bit dodgey ... so filter these ones out.
    
    After the second fetch the results are rock solid and stable
    across all platforms.

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates - github #23, qa and pmlogger review changes, Ken McDonell <=