pcp
[Top] [All Lists]

pcp updates: kenj + fche merges

To: pcp@xxxxxxxxxxx
Subject: pcp updates: kenj + fche merges
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue, 2 Apr 2013 00:39:38 -0400 (EDT)
Delivered-to: pcp@xxxxxxxxxxx
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: TDeyFB/zT9CbXdeNTcTdTQNLO3wNZQ==
Thread-topic: pcp updates: kenj + fche merges
Changes committed to git://oss.sgi.com/pcp/pcp.git dev

 Makepkgs                        |    3 +
 build/cleantmpdirs              |    5 +--
 qa/169                          |   41 ++++++++++++++++++++------
 qa/462                          |   41 +++++++++++++++++++++-----
 qa/admin/check-vm               |   17 +++++++++-
 qa/common.secure                |    2 +
 qa/qa_hosts.master              |    1 
 src/libpcp/src/context.c        |   18 ++++++++---
 src/libpcp/src/internal.h       |    1 
 src/libpcp_pmcd/src/GNUmakefile |    6 +++
 src/libpcp_pmcd/src/trace.c     |   62 +++++++++++++++++++++++++++++-----------
 src/pmcd/src/client.c           |   17 +++++++---
 src/pmcd/src/client.h           |    5 +--
 13 files changed, 168 insertions(+), 51 deletions(-)

commit 3d59a434fe20bc64b2683a90f32d0c5bfacd3d6c
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Apr 2 15:37:29 2013 +1100

    Merge Kens mv-without-minus-n change into generic tmp script too

commit 7adaa9e9f2733dc2abf832b22f4db11803a1c0c8
Merge: 1cce310 4b69271
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue Apr 2 15:33:35 2013 +1100

    Merge branch 'dev' of git://oss.sgi.com/kenj/pcp into dev

commit 1cce310be6d4036b40c29632b9b53994d23ceef0
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Fri Mar 29 13:15:20 2013 -0400

    libpcp context.c: Avoid mis-sharing outbound pmcd connections if ports 
mismatch
    
    pmNewContext tries to reuse existing connections to a pmcd, if a
    second context request comes in.  However, for this optimization, it
    only compared host names and not port numbers, which leads to a new
    flavour of false sharing.

commit 4b69271ebe5847dc730a5135a2fd1318ee8770ee
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Tue Apr 2 13:42:30 2013 +1100

    pmcd client connection tracing, qa/169
    
    Fix for failing qa/169 which involved an unintended DNS timeout
    in gethostbyaddr() at a critical point in the concurrent timeout
    handling for both a failing PMDA and a client request.
    
    In pmcd, do not do reverse DNS lookups for trace diagnostic,
    report IP addr only.  Also rework this code to handle the mix
    { secure sockets enabled, secure sockets disabled } x { ipv4, ipv6 }.
    
    Involves exposing the internals of pmcd's ClientInfo struct to
    libpcp_pmcd, and getting some of the internal socket state from
    libpcp visible in libpcp_pmcd.

commit 75ca0ce763b62d161cc19e61f14d5575d2976eac
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Mar 31 07:33:46 2013 +1100

    qa/admin - more /etc/hosts sniffing for check-vm

commit 4d52928fe7de261b22736443747949c99d160c5e
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Mar 31 07:32:21 2013 +1100

    qa/common.secure - include ascii certificate in *.full
    
    Just being paranoid in my attempts to debug the failing 71? tests
    ... but this may be useful at some point in the future.

commit 1a16f430ffde5ac14ffe69e3d021182f97e6beb0
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Mar 31 07:28:27 2013 +1100

    Makepkgs - make mini-clean substitute
    
    There are _some_ artifacts left in the workarea (especially after
    a QA run) that can break a following Makepkgs run, e.g. qa/qa_outfiles.
    
    This commit provides a place for dealing with these, without requiring
    a make clean (which obliterates things that take a long time to
    recreate the next time you run QA).

commit 65b1c6d8ba9bf98c986c667e2e95f9cc69e43a84
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Mar 31 07:27:43 2013 +1100

    qa_hosts.master - add host emma for kenj

commit fb2240c55b229e57d182478c5e8f602a17eb43e6
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Mar 31 07:23:20 2013 +1100

    qa/462 - fix nasty race condition
    
    Between tests 4 and 5 there was the possibility for a nasty
    (i.e. very hard to debug) race between the backgrounded pmcd
    stop and the following pmcd start in the foreground.
    
    Added a timed semaphore using a temporary file and improved the
    diags sent to 462.full.

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates: kenj + fche merges, Nathan Scott <=