Changes committed to git://git.pcp.io/pcp.git master
Hemant Kumar (5):
qa: add multi-node sysfs/procfs memory test data for injection
pmdalinux: update max numa bandwidth iff the .config changes
pmdalinux: additional qa for mem.numa.max_bandwidth
pmdas/perfevent: Add a perf derived event for current memory bandwidth
qa/perfevent: Changes related to derived events
Ken McDonell (4):
src/libpcp/src/logportmap.c: fix parenthesis in if ((sts = ...
qa/1092: don't trust the src/getversion binary
src/libpcp/src/win32.c: fix off-by one error in strncpy() uses
src/libpcp: small mem leak plugs
Nathan Scott (4):
Drop python dep for main pcp package
pmdalinux: small refinements to numa mem bandwith refresh logic
build: slight correction to python rpm version deps
qa: add python mmv test qa/704 into the default set
Anthony DeChiaro (2):
Python bindings: Fixing bug where name isn't set in mmv_metric
Python bindings: Fixing crash/not updating metrics via lookup_mapping
Marko Myllynen (2):
pmrep: zap shell completion todo list item
pmdaproc: fix typo in name of proc.memory.vmhw{n,m} metric
Lukas Berk (1):
RPM Packaging - new pcp-devel package with tools/docs
build/rpm/fedora.spec | 51 -
build/rpm/pcp.spec.in | 53 -
qa/022.out.linux | 2
qa/1092 | 16
qa/704 | 32
qa/704.out | 74 -
qa/821.out | 1580 ++++++++++++++++++++++++++++++++++++
qa/826 | 28
qa/826.out | 18
qa/943.out | 4
qa/group | 2
qa/linux/meminfo-root-002.tgz |binary
qa/perfevent/perf_event_test.c | 8
qa/src/test_mmv.python | 2
src/libpcp/src/auxconnect.c | 7
src/libpcp/src/logportmap.c | 2
src/libpcp/src/logutil.c | 4
src/libpcp/src/util.c | 6
src/libpcp/src/win32.c | 4
src/pmdas/linux/mem_bandwidth.c | 45 -
src/pmdas/linux/numa_meminfo.c | 3
src/pmdas/linux/numa_meminfo.h | 1
src/pmdas/linux_proc/help_text.h | 4
src/pmdas/linux_proc/pmda.c | 8
src/pmdas/linux_proc/proc_dynamic.c | 2
src/pmdas/linux_proc/proc_pid.c | 4
src/pmdas/linux_proc/proc_pid.h | 4
src/pmdas/perfevent/perfevent.conf | 10
src/pmdas/perfevent/perfinterface.c | 68 -
src/pmdas/perfevent/perfinterface.h | 49 +
src/pmrep/TODO | 1
src/python/pcp/mmv.py | 47 -
32 files changed, 1952 insertions(+), 187 deletions(-)
commit db4b39656591654d8a9a33f778f6b8f20852ab6b
Author: Marko Myllynen <myllynen@xxxxxxxxxx>
Date: Mon May 16 16:34:37 2016 +1000
pmdaproc: fix typo in name of proc.memory.vmhw{n,m} metric
From http://man7.org/linux/man-pages/man5/proc.5.html:
* VmHWM: Peak resident set size ("high water mark").
There's no proc.memory.vmhwm available as one would expect but
proc.memory.vmhwn - and it turns out that the same typo is present
everywhere in PCP sources so the value will always be zero.
commit 1fe09f69e4502fa3b97ed5d12deb8440ac65c067
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon May 16 16:25:27 2016 +1000
qa: add python mmv test qa/704 into the default set
Make test 704 deterministic and fix up a python2/3 porting issue.
commit 47ead6f30a4ac3a7f24bce1ea6d5c183189e616a
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon May 16 14:53:09 2016 +1000
src/libpcp: small mem leak plugs
All on error code paths that I'd expect to be rarely (if ever)
executed in the wild.
commit a89fa873c13070750c08ac3b2092a72f9036e1cd
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon May 16 14:52:45 2016 +1000
src/libpcp/src/win32.c: fix off-by one error in strncpy() uses
commit 655f4e18eb516be79e7eb23e834dc6475d7475eb
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon May 16 14:50:53 2016 +1000
qa/1092: don't trust the src/getversion binary
Force it to be rebuilt ... stale binary was triggering failures when
new PCP package version rolled over in QA.
commit aad6eaa42b9ec6767d8ba281c4c2d14a3ddc94b4
Author: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
Date: Mon May 16 12:53:20 2016 +1000
qa/perfevent: Changes related to derived events
Change the tests for mismatch/absent counters to check for the no. of
derived events. Only one derived event should be registered in each
case. The other derived event shouldn't be registered.
Output remains unchanged and so is 756.out.
Signed-off-by: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
commit 357140fa8575ca9a7ddf7379b5b1ad11c64943bb
Author: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
Date: Mon May 16 12:52:54 2016 +1000
pmdas/perfevent: Add a perf derived event for current memory bandwidth
Adds a derived perfevent current_bandwidth to perfevent.conf file. This
metric gives the current memory bandwidth usage for a system. Currently,
only x86 counters are added.
# pminfo | grep current_bandwidth
perfevent.derived.current_bandwidth.value
# pmval perfevent.derived.current_bandwidth.value
metric: perfevent.derived.current_bandwidth.value
host: <some_host>
semantics: cumulative counter (converting to rate)
units: count (converting to count / sec)
samples: all
cpu0 cpu8
78.47 541.0
77.66 562.1
641.4 232.2
232.7 859.4
67.59 944.5
3.719E+04 3.710E+04
4.084E+04 4.077E+04
4.042E+04 4.039E+04
1.128E+04 1.126E+04
1.421E+04 1.427E+04
...
This patch also makes some changes to code, to allow other perf derived
events to be defined even if one or more derived events fail to
register. So, if the current_derived derived event fails to register due
to unavailability of some of its base counters, we log an error to
perfevent.log and continue registering other derived events.
Some data structure definitions have been moved from perfinterface.c to
perfinterface.h to allow them to be used in qa/perfevent.
Signed-off-by: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
commit 2807404b95cf819a1346b860e2e9f2aa7076d08b
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon May 16 12:32:38 2016 +1000
build: slight correction to python rpm version deps
commit 4ecd9d4e7f642f67df4a446bde95f766cc0befe8
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon May 16 12:22:10 2016 +1000
pmdalinux: small refinements to numa mem bandwith refresh logic
Now handles the case where a config file existed then the file is
removed. Also, less chatty in the PMDA log file for the expected
situation of a conf file not yet existing.
Updated test qa/826 to match the current behaviour.
commit 13ebad831689015f446044632c12d98a729cb99c
Author: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
Date: Mon May 16 12:19:30 2016 +1000
pmdalinux: additional qa for mem.numa.max_bandwidth
Add tests for multinode configuration.
Signed-of-by: Hemant Kumar <hemant at linux.vnet.ibm.com>
commit e5bef6ffb9dea90973dd454ce2f3260a3f0a5228
Author: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
Date: Mon May 16 12:18:29 2016 +1000
pmdalinux: update max numa bandwidth iff the .config changes
Add a check for max memory bandwidth.conf file against the last modified
time and accordingly update the metric values.
Signed-off-by: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
commit 416239bf2ba4abfad63b52cf2817efd26ac8bfcb
Author: Marko Myllynen <myllynen@xxxxxxxxxx>
Date: Mon May 16 11:34:08 2016 +1000
pmrep: zap shell completion todo list item
commit 0db7cd12f0cc53c1a292d50790e2d9e41dfb02dc
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sun May 15 15:23:44 2016 +1000
src/libpcp/src/logportmap.c: fix parenthesis in if ((sts = ...
Reported as https://github.com/performancecopilot/pcp/issues/93
commit 65aa5be959f6e3962eb42657a072a60b1d9aa210
Author: Anthony DeChiaro <adechiaro@xxxxxxxxxxxx>
Date: Fri May 13 16:40:36 2016 -0700
Python bindings: Fixing crash/not updating metrics via lookup_mapping
commit c0a1a508682d3b56d5737d5de28598674e08c79d
Author: Anthony DeChiaro <adechiaro@xxxxxxxxxxxx>
Date: Fri May 13 14:08:39 2016 -0700
Python bindings: Fixing bug where name isn't set in mmv_metric
commit 1e745ab24d2e0dc1d01d25619c360cd27b39bb55
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri May 13 11:48:03 2016 -0400
Drop python dep for main pcp package
commit dc89e74e859f8b4c0a0ea0323614b716a569e3a6
Author: Lukas Berk <lberk@xxxxxxxxxx>
Date: Fri May 13 11:36:26 2016 -0400
RPM Packaging - new pcp-devel package with tools/docs
Leave pcp-libs-devel for just the libs and header files,
push devel pmdas and tools into pcp-devel and adjust the deps
accordingly
commit a51d3b8baecb6076a0b9814a546d0735993871ed
Author: Hemant Kumar <hemant@xxxxxxxxxxxxxxxxxx>
Date: Fri May 13 12:32:08 2016 +1000
qa: add multi-node sysfs/procfs memory test data for injection
Expected output from test 821 updated accordingly.
|