Nathan's on vacation, so I'm pushing these to the official tree and github.
Changes committed to git://git.pcp.io/pcp.git master
Ken McDonell (19):
qa/pmdas/github-56: add .gitignore
src/pmdas/trace: fix demos Makefile
qa/common.filter: more tweaks for _filter_pcp_start()
src/pmdas/trace: more installed makefile changes
qa/common.rc: rework recent sudo -E test change
qa/README: document that things work better if the user pcpqa is included
in the pcp group
qa/admin/pcp-daily: report date after QA done
qa/common.rc: sudo -h output sometimes on stderr ... sigh.
src/perl/PMDA/test.pl: correct a pattern match in the test program
src/libpcp/context.c: tweak pmNewContext() when re-using an existing pmcd
connection
qa/1090: (new) tests for
https://github.com/performancecopilot/pcp/issues/14
qa/group: 1090 belongs in the "context" group as well
qa/024: remake .out as new "ping" PDUs visible below pmNewContext()
qa/pmdas/broken/GNUmakefile.install: fix rules so broken pmdas are
rebuilt for QA
qa pmdas: add some temporary files (aprÃs qa) to .gitignore
qa/1090: when testing pmwebd, at least make sure it is running
qa/admin/check-vm: some more debian build package prereqs
qa/admin/pcp-daily: workaround for VM shutdown oddness
qa/pmdas/dynamic/GNUmakefile: avoid somwe qa/956 failures
qa/024.out | 2
qa/1090 | 130 ++++++++++++++++++++++++++++++++++++
qa/1090.out | 33 +++++++++
qa/374 | 5 +
qa/README | 3
qa/admin/check-vm | 17 ++--
qa/admin/pcp-daily | 5 +
qa/common.filter | 6 +
qa/common.rc | 11 ++-
qa/group | 1
qa/pmdas/broken/GNUmakefile.install | 6 -
qa/pmdas/dynamic/GNUmakefile | 37 ++++++++++
qa/pmdas/github-56/.gitignore | 2
qa/pmdas/memory_python/.gitignore | 1
qa/pmdas/slow/.gitignore | 2
qa/pmdas/slow_python/.gitignore | 1
qa/pmdas/test_perl/.gitignore | 1
src/libpcp/src/context.c | 43 +++++++++++
src/perl/PMDA/test.pl | 2
src/pmdas/trace/.gitignore | 1
src/pmdas/trace/GNUmakefile | 15 +++-
src/pmdas/trace/Makefile.proto | 2
22 files changed, 304 insertions(+), 22 deletions(-)
Details ...
commit 42d26403d6d66968f7650ef0863725ce5e50711d
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Dec 28 06:30:44 2015 +1100
qa/pmdas/dynamic/GNUmakefile: avoid somwe qa/956 failures
When running QA from a git tree (not the testsuite directory)
we try hard to use GNUmakefile.install in preference to
GNUmakefile so we're not dependent on the state of the git
tree. But qa/956 exposes a corner case when the git tree
is pristine, then the dynamic PMDA (QA only special) fails
to Install because the Install script via pmdaproc.sh runs
a "make install" that uses GNUmakefile ... tart up GNUmakefile
to do the right thing when this happens.
commit ed55f15d56436011da332558993a2b4462762bad
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Dec 28 06:26:33 2015 +1100
qa/admin/pcp-daily: workaround for VM shutdown oddness
On some of my VMs, there is a problem with the shutdown that deletes
some files (usually qa scripts from git) which causes the subsequent
Makepkgs to fail.
This happens often enough that I've decided to defend against it.
commit c7329467bab4758aad91eb74b81ec309f0c42786
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Dec 28 06:25:24 2015 +1100
qa/admin/check-vm: some more debian build package prereqs
Exposed by upgrade to LinuxMint 17.3 (vm18).
commit 00e52a974eaf186b7b6a85a897f70c98344c4adf
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Dec 28 06:22:57 2015 +1100
qa/1090: when testing pmwebd, at least make sure it is running
commit 000e2cc912539e4f61c4196fae2c2a6e9a0dca05
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Dec 28 06:20:56 2015 +1100
qa pmdas: add some temporary files (aprÃs qa) to .gitignore
commit 2276bafec95e25868995e793b6e17da0e6e53532
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Dec 23 10:10:55 2015 +1100
qa/pmdas/broken/GNUmakefile.install: fix rules so broken pmdas are rebuilt
for QA
commit 3be30eed8e08ac69a63f72c3298774ece5be7648
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Dec 23 06:36:05 2015 +1100
qa/024: remake .out as new "ping" PDUs visible below pmNewContext()
commit 84f4c396e508c23ccdbcdfda8537f1ecf2b6415c
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Dec 23 06:34:36 2015 +1100
qa/group: 1090 belongs in the "context" group as well
commit 520beefaf38bdb64d51396dadb369befd1342f7c
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Dec 23 06:32:38 2015 +1100
qa/1090: (new) tests for https://github.com/performancecopilot/pcp/issues/14
Exercise new "ping" protocol between libpcp's pmNewContext and pmcd
when re-using a socket connection for a new PMAPI context.
commit 3d4d2c07efa4954bc188420ad5ad85ebd35481dd
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed Dec 23 06:12:04 2015 +1100
src/libpcp/context.c: tweak pmNewContext() when re-using an existing pmcd
connection
There is fast path (and fd saving) optimization in pmNewContext()
where we don't establish a new connection to pmcd when we're asked
to create a subsequent PMAPI context to the same pmcd host and port
used to create an existing PMAPI context. Multiple PMAPI contexts
are then multiplexed over the same socket connection. This all
works as designed.
But as shown in https://github.com/performancecopilot/pcp/issues/14
this produces slightly odd behaviour when pmcd has died between the
initial pmNewContext() call and the later one, but the client has not
sent a PDU and so libpcp does not update the state of the connection
to pmcd and will try to re-use the socket that is not going to work.
Since we cannot rely on any socket-level service to let us know
the remote end of a socket has been closed, we need a "ping" ... we
don't have a ping PDU per se (and adding one would introduce version
compatibility issues) so we use a (bad) pmDesc request PDU with an
expected PM_ERR_PMID error PDU response ... if this happens, pmcd is
well, otherwise go down the slow path and try to build a new socket
connection to (presumably a new) pmcd.
commit 1e44d6c90dd4c99df8a6037f741d4d13a194196d
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Dec 21 17:29:30 2015 +1100
src/perl/PMDA/test.pl: correct a pattern match in the test program
Was causing qa/771 to fail on some more modern Perl versions, like
vm31 (Fedora 23). Thanks Nathan for the guidance.
commit de828f89f6654f40addfbe8b9e19ce35c4051ee7
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 17:31:40 2015 +1100
qa/common.rc: sudo -h output sometimes on stderr ... sigh.
commit ef3209c9d69cca3b6a64bc9e4239b53e744f7e40
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 12:00:31 2015 +1100
qa/admin/pcp-daily: report date after QA done
Helps managing logs when more than one QA run per day.
commit ac90e4b2a4106b577dabeebe4addb58ada8e0a28
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 11:53:59 2015 +1100
qa/README: document that things work better if the user pcpqa is included
in the pcp group
commit c66c355069cbaf5fd73c0040ae45db99ec65a7d6
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 10:49:00 2015 +1100
qa/common.rc: rework recent sudo -E test change
Recently we changed the way sudo was run to test for the presence
of a -E option (to preserve the environment) ... this was not quite
right for some versions of sudo, triggering a batch of new QA failures.
commit 7e6a5304a8133615b40c901bc2eb1cf09b51f7a5
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 10:45:21 2015 +1100
src/pmdas/trace: more installed makefile changes
Need to do some filtering in the build to avoid undefined macros
when installed ... was masked previously because sudo was typically
run with -E from the qa check script (botching the recent sudo -E
check in qa/common.rc exposed this problem).
commit ec1ed38f9c5ab7452ecf7516f1b451395836f3c2
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 06:42:05 2015 +1100
qa/common.filter: more tweaks for _filter_pcp_start()
On grundy I've started to see "..done" being appended to the
Starting pmcd ... and Starting pmlogger ... lines.
Remove "..done" in the filter.
commit 7ad7f900531d8bfd47512c2e3c31fe88f42bd077
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 06:36:45 2015 +1100
src/pmdas/trace: fix demos Makefile
I don't understand how qa/258 was working on most hosts ... the
installed Makefile had no chance of building any of the app? apps if
they did not already exist.
commit 82daaac794c093505cae7249796326c734fbafc3
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Sat Dec 19 06:17:51 2015 +1100
qa/pmdas/github-56: add .gitignore
|