Changes committed to git://git.pcp.io/pcp.git master
Nathan Scott (6):
docs: link the INSTALL file from README properly
qa: ensure 878 cleans up, else post-test checks can fail
qa: resolve pmwebd -I-induced regression in test qa/661
qa: rework python-not-configured checks for tests
qa: make python interpreter diagnostic more verbose
qa: fix a cut+paste error in test 1024 data
README.md | 2
qa/056 | 5
qa/073 | 5
qa/1024.out | 289 ++++++++++++++++++++++++++++++++++++-------
qa/1052 | 4
qa/1053 | 4
qa/1062 | 4
qa/1068 | 4
qa/1069 | 4
qa/1113 | 5
qa/547 | 3
qa/553 | 4
qa/661 | 4
qa/667 | 5
qa/702 | 5
qa/704 | 5
qa/707 | 5
qa/708 | 5
qa/709 | 5
qa/710 | 5
qa/717 | 5
qa/718 | 5
qa/722 | 6
qa/729 | 11 -
qa/737 | 4
qa/739 | 5
qa/741 | 5
qa/742 | 5
qa/743 | 5
qa/784 | 5
qa/829 | 5
qa/839 | 5
qa/842 | 5
qa/843 | 5
qa/859 | 5
qa/878 | 1
qa/880 | 6
qa/904 | 4
qa/949 | 5
qa/968 | 4
qa/979 | 5
qa/980 | 5
qa/990 | 21 +--
qa/991 | 5
qa/995 | 4
qa/GNUmakefile | 6
qa/common.python | 4
qa/linux/blkdev-root-003.tgz |binary
48 files changed, 293 insertions(+), 230 deletions(-)
commit 62dcbf8bbc8d2218335a34497355a80774c0bbd2
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Wed Jun 8 13:38:01 2016 +1000
qa: fix a cut+paste error in test 1024 data
One of the procfs files used to inject testing data into
pmdalinux for block metric testing had a cut+paste error!
Turns out that on big endian platforms, this induces some
indeterminism in the output which doesn't exist otherwise
and the test fails.
Since the kernel will never produce dodgey data of that
nature, not worth defending against in pmdalinux - simply
fix the sample data.
commit 4359fbef89aaa6fb96c56e76872fb6f9a0ae6b5f
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Wed Jun 8 11:26:01 2016 +1000
qa: make python interpreter diagnostic more verbose
commit 8ccfb402df40bd7c8a46681608d6b18eef706140
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Wed Jun 8 11:10:31 2016 +1000
qa: rework python-not-configured checks for tests
Found several tests with python dependencies that we were
not checking for common.python with the original scheme.
Now, we unconditionally include common.python (instead of
compile-time conditional) so tests can always use it (and
are simpler as a result), and use the _notrun check there
for PCP_PYTHON_PROG as the main test guard.
commit d84c22e71535349f335e9e959d7ec536e5035722
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Jun 7 16:41:48 2016 +1000
qa: resolve pmwebd -I-induced regression in test qa/661
Seeing pmwebd test 661 failure on all Red Hat QE hosts, git
bisect finds it to be a regression from e9262b89b0fe where
a -I was added into pmwebd test options. This changes the
way the -A option is handled, and is causing the two metric
regex parts of the test to no longer match on their archive
name components, so curl ends up producing no output.
While in there, also fix a small race where pid might not be
set by the time _cleanup is called, causing kill invocation
on an empty $pid.
commit 1b8db157f8d3afb6b7ed383560438a8da590482c
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Jun 7 10:12:27 2016 +1000
qa: ensure 878 cleans up, else post-test checks can fail
commit 63d4bb45d0a108729c44cfd26d4db0c658c8650f
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Jun 7 09:36:21 2016 +1000
docs: link the INSTALL file from README properly
|