Changes committed to git://git.performancecopilot.org/pcp.git dev
qa/729 | 48 +++++++++++++++++++++++++++++++++++++++
qa/729.out | 5 ++++
qa/group | 1
src/pmcollectl/pmcollectl.py | 16 +++++++------
src/pmdas/pmcd/src/pmcd.c | 2 -
src/pmdas/postfix/pmdapostfix.pl | 4 +--
src/python/pcp/pmapi.py | 7 ++++-
src/python/pcp/pmcc.py | 14 +++++++----
8 files changed, 80 insertions(+), 17 deletions(-)
commit b361623aaa15998f15f27096c731478ef06b2246
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue May 27 12:45:56 2014 +1000
Small pcp.pmcc module fixups, esp for archive mode operation
Remove unused imports from pcp.pmcc and ensure we call archive
variant of indom lookup API. This moves us a little closer to
passing test qa/980 - much work remains, however.
commit edc074f1c46e8f3e2bc78a22a6410de2af09f8f7
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue May 27 12:39:33 2014 +1000
pmcollectl: fix sub-second recording mode QA anomaly
A misunderstanding on my part on how the odd-looking +1sec
to all pmcollectl recordings was supposed to do something
useful, resulted in pmlogger popping up those pesky GUI-mode
record dialogs after pmcollectl recording.
Fix this in a way that will remove them for good while also
allowing short recording sessions to work.
commit 88ad1739e3e4129000a8730ee6ecbafd0cdaf367
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue May 27 10:50:28 2014 +1000
postfix pmda: run as the postfix user, not the pcp user
The Postfix PMDA uses qshape(1) to gather postfix queue stats,
by reading files the various postfix queue directories. All
of the postfix queues are only readable by the postfix user or
root.
Initial problem report and testing brought to you by the good
folks at messagemedia.com.au.
commit 592f2520ff7b984a907afc91f090edb2cab5c8ec
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon May 26 10:41:47 2014 +1000
Automate manual test cases from recent python API fixes
commit 7c8fbda3c4a64720df7d8ace71e8e2e6843ce574
Author: Michele Baldessari <michele@xxxxxxxxxx>
Date: Sat May 24 22:05:29 2014 +0100
Fix python pmParseInterval failure segfault
The following testcase segfaults because errmsg is NULL:
"""
import pcp.pmapi
context = pcp.pmapi.pmContext(target='local:')
print(context.pmParseInterval(''))
"""
Program received signal SIGSEGV, Segmentation fault.
0x0000003d2e43182d in parseError (spec=spec@entry=0x3d2e4591c6 "",
point=point@entry=0x3d2e4591c6 "", msg=msg@entry=0x3d2e45b166 "Null or empty
specification", rslt=rslt@entry=0x0) at rtime.c:202
202 if ((*rslt = malloc(need)) == NULL)
(gdb) bt
Make pmParseInterval parameters the same as pmParseMetricSpec()
commit 41142f633ffe8c9d75226b15bca8d99faa10dc39
Author: Michele Baldessari <michele@xxxxxxxxxx>
Date: Sat May 24 11:00:55 2014 +0100
Python - Fix segfault when printing empty units
The following example:
"""
import pcp.pmapi
from pcp.pmapi import c_api
fname = '20140506.0'
context = pcp.pmapi.pmContext(c_api.PM_CONTEXT_ARCHIVE, fname)
pmids = context.pmLookupName(['proc.nprocs'])
descs = context.pmLookupDescs(pmids[0])
print 'Segfault: %s' % descs[0].contents.units
"""
will segfault:
pmUnitsStr_r (pu=0x0, buf=buf@entry=0x3d2e66f9c0 <ubuf> "",
buflen=buflen@entry=60) at units.c:177
177 if (pu->dimSpace) {
Make sure we wrap pmUnitsStr in order to catch these and correctly
return an empty string
commit c8bf7a1afda58129dbe0747b5f86da7dadd43269
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Sat May 24 11:48:55 2014 +1000
Fix pmdapmcd off-by-one preventing 3 or more pmcd.services
|