Changes committed to git://git.pcp.io/pcp.git master
Ken McDonell (11):
qa/1071: chmod to be executable
qa/common.config: wrestle with X11 DISPLAY setting in the QA Farm
qa/admin/pcp-daily: tweak warnings filter
src/include/builddefs.in: fixes for python packaging
qa/admin/check-vm: adjust specs for python-requests
qa/common.config: more wrestling with X11 DISPLAY setting in the QA Farm
various: indentation and white space changes
qa/common.config: undo DISPLAY voodoo for some vm* hosts
Makepkgs & qa/admin/myconfigure: disable Python bits for FreeBSD 9.3
tar&mac installs: handle dir renaming below $PCP_VAR_DIR/config/pmieconf
qa/common.rc: big hammer for LC_COLLATE
Dave Brolley (9):
New PMAPI calls: pmSemStr(3) and pmSemStr_r(3).
pmlogger: Validate all configured metrics after PMCD_ADD_AGENT is
received from pmcd.
New man page for pmSemStr(3) and pmSemStr_r(3).
Python binding for pmSemStr(3).
New qa test 891 and python testing for pmSemStr(3) and pmSemStr_r(3).
libpcp: Move pmSemStr(3) and pmSemStr_r(3) to the next libpcp version.
Replace stubs for qa tests 891 932 with the actual tests.
PCP Bug 1100: Validate metrics before generating MARK record.
New qa for Bug 1100.
Nathan Scott (8):
qa: additional sample programs for aspects of MMV format
qa: use suggestion from mjw for older valgrind warning suppression
pmdaproc: rewrite proc.psinfo.ttyname metric handling
qa: fix some typos and some dropped qa/group updates
pmproxy: make probe-style connection error a debug diagnostic
docs: note alternate sources of mark records in pmdumplog(1)
docs: update pmmgr man page to follow pcp conventions
pmlogger: auto-reconnect to pmcd whenever possible
Makepkgs | 17 +
build/mac/installer-resources/postinstall | 12
build/tar/postinstall.tail | 14
man/man1/pmdumplog.1 | 7
man/man1/pmmgr.1 | 401 ++++++++++++++-----------
man/man3/pmidstr.3 | 3
man/man3/pmindomstr.3 | 3
man/man3/pmsemstr.3 | 98 ++++++
man/man3/pmtypestr.3 | 5
man/man3/pmunitsstr.3 | 8
qa/1055 | 133 ++++++++
qa/1055.out | 24 +
qa/389 | 21 -
qa/389.out | 18 -
qa/854 | 133 ++++++++
qa/854.out | 24 +
qa/856 | 133 ++++++++
qa/856.out | 24 +
qa/882 | 134 ++++++++
qa/882.out | 24 +
qa/891 | 42 +-
qa/891.out | 8
qa/932 | 115 ++++++-
qa/932.out | 21 +
qa/admin/check-vm | 4
qa/admin/myconfigure | 17 +
qa/admin/pcp-daily | 7
qa/common.check | 7
qa/common.config | 16 -
qa/common.rc | 6
qa/group | 12
qa/src/.gitignore | 3
qa/src/GNUlocaldefs | 5
qa/src/mmv_ondisk.c | 113 +++++++
qa/src/mmv_simple.c | 35 ++
qa/src/semstr.c | 69 ++++
qa/src/test_pcp.python | 6
src/include/builddefs.in | 6
src/include/pcp/pmapi.h | 2
src/libpcp/src/exports | 7
src/libpcp/src/getopt.c | 121 +++----
src/libpcp/src/util.c | 50 ++-
src/libpcp_pmda/src/open.c | 3
src/newhelp/chkhelp.c | 6
src/pcp/atop/showlinux.c | 22 -
src/pmchart/view.cpp | 9
src/pmdas/apache/apache.c | 4
src/pmdas/bash/bash.c | 4
src/pmdas/linux_proc/getinfo.c | 67 ++--
src/pmdas/linux_proc/getinfo.h | 4
src/pmdas/linux_proc/pmda.c | 6
src/pmdas/lmsensors/lmsensors.c | 468 +++++++++++++++---------------
src/pmdas/weblog/weblog.c | 5
src/pmlogger/src/check.c | 143 ++++++++-
src/pmlogger/src/fetch.c | 37 +-
src/pmlogger/src/logger.h | 6
src/pmlogger/src/pmlogger.c | 52 ++-
src/pmproxy/client.c | 21 -
src/python/pcp/pmapi.py | 12
59 files changed, 2122 insertions(+), 655 deletions(-)
commit 695da0b90ff32e53c080e7666d07cac1f3065af1
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 24 15:01:24 2016 +1000
pmlogger: auto-reconnect to pmcd whenever possible
Extend Dave's consistency checking to also allow pmlogger
to reconnect automatically to pmcd (just like pmie always
has).
This is important for the common case of a pmcd restart on
a host monitored by a central pmlogger farm. Previously,
we were open to some minutes of missed data until the cron
pmlogger_check kicked in - now it is instantaneous.
Its worth noting that we still always start a new pmlogger
with each new day, even with this change in place.
Verify the auto-reconnection and presence of a mark record
via the qa/389 test script.
commit 9613f182e9df099011a0531365b8a3df69119b8c
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 24 14:47:25 2016 +1000
docs: update pmmgr man page to follow pcp conventions
commit 1a3c9bd752bb5c94936e3a25e8de821e380e1f7b
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 24 14:45:58 2016 +1000
docs: note alternate sources of mark records in pmdumplog(1)
commit 66d8a0c647008c4a960c013c3e4e8bcb71af2e8f
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 24 09:48:04 2016 +1000
pmproxy: make probe-style connection error a debug diagnostic
commit ca08dbb45d4c75cf57492e4043367a908090c9e4
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Fri Jun 24 09:44:24 2016 +1000
qa: fix some typos and some dropped qa/group updates
commit 1b5e787b147629ac72aff8912c36d748a85195b1
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Jun 23 11:07:55 2016 -0400
New qa for Bug 1100.
854, 856, 882, 932 1055.
commit 4399c94dbb4f2ff485f1dddc4258b0efa34506c8
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Jun 23 11:06:23 2016 -0400
PCP Bug 1100: Validate metrics before generating MARK record.
commit bf69c680113c1199ae3e8db5d6efa0c9b872ad51
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Jun 23 19:59:33 2016 +1000
qa/common.rc: big hammer for LC_COLLATE
It turns out that on some platforms the RE [A-Z] really means [A-Za-z]
by default ... we use this all over in QA. While this (based on how
long has been lurking) is mostly benign, it does bite in some places,
like qa/1071 that will fail on some platforms but only in June or July
... no, I'm not joking!
Setting LC_COLLATE=C in the environment enforces the original Unix
semantics that was prevailing when most of the QA tests were written
and returns us to a sane place.
commit 10f689e46dceaaba5d9b8f2a24bdfbccaa4d1aff
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Jun 23 19:57:44 2016 +1000
tar&mac installs: handle dir renaming below $PCP_VAR_DIR/config/pmieconf
$PCP_VAR_DIR/config/pmieconf/pcp should not exist, it has moved to
$PCP_VAR_DIR/config/pmieconf/primary
commit cd39b0a43f455be0e4ae6c596f196623c1e82ab7
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 23 17:41:58 2016 +1000
pmdaproc: rewrite proc.psinfo.ttyname metric handling
Resolves some short-comings in proc.psinfo.ttyname
- stat'ing and realpath'ing every open fd when a dev match
not found can be expensive;
- it also generates many SElinux warnings, since the fds
can point anywhere (or maybe sockets, fifo's, etc);
Restrict the search to just the first three (standard) fds
(even though these may be remapped) since its not likely
we will find a match if not here. Use readlink on the proc
files instead of stat to avoid following the link, keeping
SElinux happy. Only stat links that begin with /dev/ (the
code assumed this path previously anyway, so check for that
before making any further syscall).
Related to RH BZ #1336211.
commit bc362c9bd7ff7d99ce0e0345b40b96f532667bab
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 23 17:34:51 2016 +1000
qa: use suggestion from mjw for older valgrind warning suppression
commit 16c880df06cb9011b5422ee867dde240f6a47542
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu Jun 23 14:03:02 2016 +1000
qa: additional sample programs for aspects of MMV format
commit afa65bbd631304228c2d3359bd324a0613bd328a
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Jun 23 07:02:31 2016 +1000
Makepkgs & qa/admin/myconfigure: disable Python bits for FreeBSD 9.3
I am tired of wasting time trying to get this to work ... setup.py
and my FreeBSD 9.3 (now old, but still supported) just will not work
together, so I'm disabling the PCP Python bits for just this platform.
vm10 will rejoin the QA status reports next time.
commit b183520a1e409dbe152295a79000489b7e97d2ff
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Jun 23 06:58:11 2016 +1000
qa/common.config: undo DISPLAY voodoo for some vm* hosts
Found real fix (was bozo's X11 server setup via lightdm's config that
needed some TLC after the Ubuntu 16.04 install).
commit 9808b816cba2547374810f2068e76f13990e3718
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Jun 22 18:21:24 2016 +1000
various: indentation and white space changes
Fedora 24 has a new C compiler or has enabled new warnings,
specifically -Wmisleading-indentation which triggered a bunch of
new warnings.
Many of these are cosmetic changes, although the code is more readable
and less confusing after the changes, e.g. getopt.c.
But a couple were masking real (albeit mostly benign) bugs,
e.g. pmdaRehash() in libpcp_pmda's open.c and chkhelp.c
commit 0e89707ff0df2c779ad042129ce3c174946b3171
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Jun 22 18:20:22 2016 +1000
qa/common.config: more wrestling with X11 DISPLAY setting in the QA Farm
commit 6122d7e8a815f3d8e3254312cdfde3680368e18f
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Jun 22 18:19:05 2016 +1000
qa/admin/check-vm: adjust specs for python-requests
After Fedora 24 installation.
commit 37aeecc68f8bd7d68294e59baa6c17029a68db20
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Jun 22 18:15:39 2016 +1000
src/include/builddefs.in: fixes for python packaging
1. RE used for substitution was wrong (in a benign sort of way)
2. python3 rule included duplicate artefacts in manifest for rpm builds
commit 95a63f6ed9bf552084e6b58733750576f40ccaab
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Jun 22 18:14:55 2016 +1000
qa/admin/pcp-daily: tweak warnings filter
commit 74630dc33032d178fef096b4fb6cf697533ea0d1
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Jun 21 20:03:19 2016 +1000
qa/common.config: wrestle with X11 DISPLAY setting in the QA Farm
commit 9063137807b30122ca784049ebc30e5bd0c63d3b
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Jun 21 20:02:35 2016 +1000
qa/1071: chmod to be executable
commit f08038d8b7506a45f2fd21f043ac731dcf4e0c11
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Mon Jun 20 10:42:08 2016 -0400
Replace stubs for qa tests 891 932 with the actual tests.
commit 247ef16d3cc841a88506928e6894917a2966c939
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Fri Jun 17 14:59:15 2016 -0400
libpcp: Move pmSemStr(3) and pmSemStr_r(3) to the next libpcp version.
commit e411d2d9e0e8334c0d9af541cd8a8ef894f16058
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Fri Jun 17 11:56:08 2016 -0400
New qa test 891 and python testing for pmSemStr(3) and pmSemStr_r(3).
commit 68a68d47163e0c4d280908ff1e55eeb9721d67e6
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Fri Jun 17 11:55:06 2016 -0400
Python binding for pmSemStr(3).
commit 32c648838b1e005cc9b944383552a65ce09f5d80
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Wed Jun 15 16:53:07 2016 -0400
New man page for pmSemStr(3) and pmSemStr_r(3).
Also update pages for pmIDStr(3), pmTypeStr(3), pmInDomStr(3),
and pmUnitsStr(3) so that they all reference one another.
commit 6d2be8f73de483621779356c87180ced47585e39
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Fri Jun 10 11:05:26 2016 -0400
pmlogger: Validate all configured metrics after PMCD_ADD_AGENT is received
from pmcd.
PCP Bug 1100.
commit 7dcd9ad2e5d10eda289c5b5d681dc99687ed2319
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Fri Jun 10 11:00:24 2016 -0400
New PMAPI calls: pmSemStr(3) and pmSemStr_r(3).
Similar to pmIDStr(3), pmIDStr_r(3), pmInDomStr(3) and pmInDomStr_r(3).
|