Changes committed to git://oss.sgi.com/nathans/pcp.git dev
qa/062 | 2
qa/131 | 3
qa/243 | 6
qa/244 | 11 -
qa/273 | 15 ++
qa/430 | 16 --
qa/430-2.2post.out | 247 ----------------------------------------
qa/430-pre2.2.out | 175 ----------------------------
qa/430.3.out | 252 -----------------------------------------
qa/430.out.1 | 247 ++++++++++++++++++++++++++++++++++++++++
qa/430.out.2 | 252 +++++++++++++++++++++++++++++++++++++++++
qa/470 | 5
qa/472 | 5
qa/473 | 5
qa/474 | 5
qa/475 | 5
qa/common.config | 4
qa/qa_hosts.master | 1
src/include/pcp/impl.h | 4
src/libpcp/src/auxconnect.c | 10 -
src/perl/PMDA/PMDA.pm | 6
src/pmcd/src/pmcd.c | 6
src/pmdas/simple/pmdasimple.pl | 2
src/pmproxy/pmproxy.c | 6
24 files changed, 565 insertions(+), 725 deletions(-)
commit 6514e69ceda27dddcd52165160b5b731ee56022e
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Mar 5 14:33:47 2013 +1100
Multi-thread tests now cleanup their /tmp/func files
commit 916ca3a58cda5962999e6e5bd4caa8620039090f
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Mar 5 14:32:46 2013 +1100
Rework test 430 out file handling for consistency with other tests
In other news, this test now no longer fails. I don't get it, but
I suspect there must be some whacky indeterminism in the pmlogger
killing depending on chosen pid?. Makes no sense, but part 5 of
this test failed several times yesterday, and today it passes every
time. Will keep an eye on it.
commit 4deafa6f1a3254e75f6df2cf8ecac4c99c0f6e40
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Mar 5 11:55:45 2013 +1100
Fix recent networking-related regression uncovered by test 131
With the introduction of getaddrinfo and looping over the set of
addresses it returns we were inadvertantly clobbering the return
codes and forcing ECONNREFUSED to always be returned. Instead,
we now go with the return code of the last attempt, since it's
entirely plausible they are all failing for the same reason and
that there's handy diagnostic clues being thrown away. This was
the case in qa test 131, which exercises the connection timeout
logic.
Nearly missed this one thanks to a series of unfortunate QA host
reconfigurations, esp with Mac OS X hosts sometimes going into a
deep sleep and not even responding to ssh logins anymore (ugh).
commit 67b934f9ec7389da41bc0ad410bc7b8e9085e279
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue Mar 5 11:32:23 2013 +1100
Update QA test 273 to handle getaddrinfo looping over addresses
In changes recently made for IPv6 support we are iterating over
potentially a number of addresses, possibly both IPv4 and IPv6.
This causes the test to fail if the first address is not the one
used for the connection, as additional diagnostics are produced.
Handle this situation by "squashing" repeated occurrences of the
specific diagnostic in question into just one occurrence, which
matches the original output and intent.
commit 66699df766b4e1a879c39ac043d771c236d2947d
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 11:32:27 2013 +1100
Fix the build for native sockets, local addr var re-defined
commit 27487005ea1e057f58a68a1d2e98cca633f38d71
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 11:15:32 2013 +1100
Add a QA host thats reappeared into my local mix
commit 070e0289b15057efde99c0e218ac82c716b483e7
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 11:14:45 2013 +1100
Revert "QA config change, one of my earlier qa hosts has gone away"
This reverts commit 9400df353e2be3bdcbfee20f3cf5797833052a60.
Aaaaand its back.
commit 4fc887d8f9b5f7de6702265fdca57b84f2b412c6
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:31:00 2013 +1100
Remove pmcd/pmproxy warning introduced in recent changes
QA test 283 picked up the fact that we'd begun to attempt
to use additional ports unexpectedly. Missed taking into
account ports added via the command line.
commit 5515e7217ca17e1ed4cde10275c79633b7a7aaba
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:27:44 2013 +1100
Small perl module cleanups - missing macros, unused vars
commit b450d4663bc3e7ecc42d10ed61870ad6204d8d68
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:26:35 2013 +1100
Use explicit (void) argument for a couple of impl.h routines
Evidently compiling with impl.h in a python wrapper (with the
default setup.py-generated makefile) gives compiler warnings.
commit 9400df353e2be3bdcbfee20f3cf5797833052a60
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:24:40 2013 +1100
QA config change, one of my earlier qa hosts has gone away
commit 7767d7fb96e7f73acbfdefd5a95401d82e1789e3
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:23:53 2013 +1100
Fix ipv6 filtering in test 244, cleanup existing sed invocation
commit 50c7dc96e23e1d63b230bcaaf6a2ead88abc948e
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:22:57 2013 +1100
Revise optional ipv6 filtering logic in test 243
commit 80d6c8b3998d099bde8859ad70cdeba4e8f7eeda
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:22:09 2013 +1100
Add a notrun check to 131 is no faraway pmcd host is set
commit c9d9c5275d671faef9e80b630b36fbb705958770
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Mar 4 10:21:38 2013 +1100
Fix typo in revised 062 .out file link setup logic
|