Changes committed to git://oss.sgi.com/pcp/pcp.git dev
qa/546 | 2
qa/635 | 8 -
src/libpcp/src/access.c | 8 -
src/perl/PMDA/PMDA.xs | 7 -
src/pmcd/pmdaproc.sh | 20 +--
src/pmcd/src/config.c | 9 +
src/pmdas/simple/GNUmakefile | 2
src/pmdas/simple/Install | 2
src/pmdas/simple/pmdasimple.perl | 157 ++++++++++++++++++++++++
src/pmdas/simple/pmdasimple.pl | 157 ------------------------
src/pmdas/simple/pmdasimple.py | 241 -------------------------------------
src/pmdas/simple/pmdasimple.python | 231 +++++++++++++++++++++++++++++++++++
src/pmdas/trace/src/client.c | 5
src/pmdas/trace/src/comms.c | 33 +++--
src/python/pcp/pmda.py | 95 +++++++++-----
src/python/pmda.c | 97 +++++++++++---
16 files changed, 591 insertions(+), 483 deletions(-)
commit 590cd10136af13d7deedd0e4504db63068ac0978
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 14 19:39:46 2013 +1000
Fix an incorrect domain name from being generated by pmdasimple perl script
commit 43e1f62c18d5467e50d5ea0a9257d6292782120f
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 14 16:56:57 2013 +1000
Additional (big!) steps toward functioning python PMDAs
This took alot of investigation and searching for alternatives. :|
Thought I had it solved yesterday, but turned out the underlying C
code (python/pmda.c) was not quite responding in the way I thought.
Anyway, we now allocate temporary buffers in the python code, and
cast to the correct types (pmdaIndom/pmdaMetric arrays) before
calling into cpmda module code. Once there, we must decode these
objects differently using the Py_buffer API. Hairy, hairy stuff.
Also allow any of .pl, .perl, .py, .python as potential script name
suffixes as part of the install process.
commit 0f3ffb14f1e7641a8d76fc703f9cbd82ae14858a
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 13 19:15:11 2013 +1000
Further strides toward a working python PMDA wrapper and pmdasimple.py
Implemented code dealing with both forms of instance domain - list based
(C array inst/name pairs) or dictionary based (using pmdaCache APIs).
Work on the way the dispatch structure is populated; so far managing to
avoid expanding it in python code, since its contains unions of sets of
function pointers amongst other things - all a bit hairy to represent
neatly in python.
Fixed some incorrect @staticmethod callback-setter-method annotations.
commit 563e63ef3921ffa600abf808808ffe144e978429
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 13 13:31:59 2013 +1000
Update test 635 to ensure netstat reported interfaces match ours
It was noticed that on a host where an interface was marked down,
pminfo and netstat output do not match, because netstat will only
report on these interfaces when the -a option is presented. We
export /proc/net/dev interfaces without that filtering (client is
able to make use of network.interfaces.up to cull the riff-raff).
Easily resolved by adding -a option to netstat now that the issue
is understood.
commit b54e199b7094399df0f5eaad9960b3057f9af7d2
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 13 11:43:01 2013 +1000
Mark as experimental the user/group ACLs (via ifdef)
Temporarily marking the account-based access control extensions
to pmcd as experimental, and compile out pmcd.conf support by
default. Aim is to allow a bug-fix release to be made from the
current dev branch without extensive branch surgery. Similarly
mark qa/546 as notrun on earlier releases than the next one.
commit e9805ec7f2acfd31c304bac76e7d748af36857ec
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 13 11:28:08 2013 +1000
Fix a few socket address string memory leaks in diagnostics
|