pcp
[Top] [All Lists]

pcp updates - qa

To: pcp@xxxxxxxxxxx
Subject: pcp updates - qa
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon, 08 Sep 2014 17:31:30 +1000
Delivered-to: pcp@xxxxxxxxxxx
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.0
Changes committed to git://git.performancecopilot.org/kenj/pcp.git dev

 qa/755                     |    4 +
 qa/828                     |   37 ++++++++-
 qa/833                     |    9 ++
 qa/833.out                 |  174 ++++++++++++++++++++++-----------------------
 qa/README                  |    6 +
 qa/admin/allow-pmlc-access |   97 +++++++++++++++++++++++++
 qa/admin/check-vm          |   70 ++++++++++++++++--
 qa/admin/pcp-qa-summary    |   14 +++
 qa/common.filter           |    7 +
 qa/src/churnctx.c          |    4 -
 10 files changed, 319 insertions(+), 103 deletions(-)

commit 6aa4b68971a8110d3a17cbcad17def50a8aaaeb7
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 17:30:31 2014 +1000

    qa/README - add note about firewalls and QA needs

commit d53fb4a5e77fab86779ca088221dce8213467598
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 17:30:00 2014 +1000

    qa/755 - add short sleep to try and dodge slow start race

commit a6cfb83dc4e64abe88c3015feb1ab79aa8b96052
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:43:08 2014 +1000

    qa/common.filter - oops, fix last sed botch

commit 5ec7a388b2950d30592f0386d278493f72307e3a
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:39:14 2014 +1000

    qa/common.filter - another dbus error message to dodge

commit d9e6c8b44c114952d92e1895c55cee996d193bc2
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:34:35 2014 +1000

    qa/admin/check-vm - use same logic as allow-pmlc-access for IPv4 addr of 
local network

commit d4c5b5be3ee7cbc765fb57c3b7d6bd510a73b026
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:33:55 2014 +1000

    qa/admin/pcp-qa-summary - add header lines every 20 lines of output
    
    Too many failures and I lose track in the failure map.

commit ddfea4e6e745108097d5c661ef225c4b1e0232c6
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:32:39 2014 +1000

    qa/common.filter - tweak _filter_views()
    
    When QA is run from cron, dbus gets in the way even more, so add extra
    filtering.

commit b889c33eac2e4bbd84675aa57a8bf4241afda4fc
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:31:30 2014 +1000

    qa/828 - more valgrind mumbo jumbo
    
    On some platforms, the call stack ends at memalign <- valloc, rather
    than at malloc.  Add alternative suppressions for these ones.

commit 79fea378a974bff81a5d46bf988e2ee36a4afe3f
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Mon Sep 8 10:28:35 2014 +1000

    qa/admin/allow-pmlc-access - QA Helper
    
    For distributed QA to work, we need each QA machine to be able to
    run pmlc to connect to the primary pmlogger on each other QA machine.
    
    This script amends the config.default pmlogger configuration file to
    allow unlimited pmlc access from hosts on the same IPv4 network.
    
    Works for me in Carnegie ... others may need to extend to suit their
    local needs.

commit 106be1030e71fe6d811d722d9efe22e739fac850
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Sep 7 21:43:49 2014 +1000

    qa/833 & qa/churnctx.c
    
    Similar changes to 822.  PDU buffer pool usage is not quite
    deterministic, so relax filtering and be a little more relaxed
    in churnctx before reporting an apparent leak.

commit 49b118ebbb9a005f61bc101f5b4aeddc88c372e3
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Sep 7 21:34:09 2014 +1000

    qa/828 - bad valgrind tap dance
    
    Appears cause is not gcc version, nor valgrind version, but 32-bit
    CPU (i586 or i686 so far).

commit 22bc125d947de5e5062a772a2b81df1d07120ecb
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Sun Sep 7 21:17:00 2014 +1000

    qa/828 - more valgrind suppression tweaks
    
    paranoidLogRead() may not be inlined, and then appears as a
    additonal call frame between __pmGetArchiveEnd() and __pmLogRead().

<Prev in Thread] Current Thread [Next in Thread>