pcp
[Top] [All Lists]

pcp updates: misc small fixes, mostly python

To: pcp developers <pcp@xxxxxxxxxxx>
Subject: pcp updates: misc small fixes, mostly python
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue, 6 Oct 2015 01:24:55 -0400 (EDT)
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1174162033.49510530.1444109063429.JavaMail.zimbra@xxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: N2QiOFWjN8prkUOE84gsU7LGfxZ2HA==
Thread-topic: pcp updates: misc small fixes, mostly python
Changes committed to git://git.pcp.io/nathans/pcp.git master

Nathan Scott (5):
      python api: add simple debugging interface to access pmDebug
      pcp-atop: default pagesize/hertz values for failed fetches
      build: update spin-rawhide to gulp-build for vector-git
      pmdanfsclient: update comparison to handle string/int
      pcp2graphite: be consistent with import handling


 qa/667                               |    6 +++---
 scripts/spin-rawhide                 |   28 ++++++++++++++++++----------
 src/pcp/atop/various.c               |    6 ++++--
 src/pcp2graphite/pcp2graphite.py     |   32 ++++++++++++--------------------
 src/pmdas/nfsclient/pmdanfsclient.pl |    3 +--
 src/python/pcp/pmapi.py              |   10 ++++++++++
 src/python/pmapi.c                   |    8 ++++++++
 7 files changed, 56 insertions(+), 37 deletions(-)


commit 0ec7536b198da88c7f4a6c22ef2d85bbb8fe88ed
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Oct 6 16:22:10 2015 +1100

    pcp2graphite: be consistent with import handling
    
    Noticed some inconsistencies in handling of module imports
    in pcp2graphite, particularly cpmapi/c_api.  Probably just
    atrophied over time; picked one and used it throughout.

commit b926064a0845f3d42666da2dbd1bee1c6f96b746
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Oct 6 16:15:22 2015 +1100

    pmdanfsclient: update comparison to handle string/int
    
    Martins reported an string-to-integer comparison exception
    from pmdanfsclient, resulting from earlier changes to send
    integer error codes back to clients correctly.  QA actually
    tests this case so not 100% sure this will fix the problem,
    but lets give it a crack.

commit c6b5cca3d8dac6a6ba7f6e38daf9dfb7bb74e5ac
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Oct 6 16:12:31 2015 +1100

    build: update spin-rawhide to gulp-build for vector-git

commit 8d84c3ccb8c36fad48c0b943b33218c14d18b876
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Oct 6 16:10:28 2015 +1100

    pcp-atop: default pagesize/hertz values for failed fetches
    
    If these metrics are not available (eg not logged), we'll
    take a sigfpe due to division by zero - provide a default
    value for each in this case (so we can fail more cleanly,
    later on, most likely)... tackles one part of a problem
    Ken's discovered running QA.

commit c541b99debddef7e7cba4ca4419d4905b8b4bd75
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Oct 6 16:07:31 2015 +1100

    python api: add simple debugging interface to access pmDebug
    
    Add a read-only interface for checking libpcp pmDebug state into
    the python API.  So far, just export constants for APPL0,APPL1,
    and APPL2 flags since thats all python scripts should ever need.
    
    Update pcp2graphite to handle -D option more like the C tools -
    instead of overriding -D such that any use enables local debug,
    it now handles pmDebug setting a bit more like nature intended
    (e.g. -D PDU,APPL0 and/or just -DPDU).
    
    While in there, replaced "hack to include some explanatory text"
    with use of the interface designed for that purpose.
    
    Updated test qa/667 to test it out.

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates: misc small fixes, mostly python, Nathan Scott <=