pcp
[Top] [All Lists]

pcp updates: systemd fix, signal fixes, pmchart long options, qa

To: pcp@xxxxxxxxxxx
Subject: pcp updates: systemd fix, signal fixes, pmchart long options, qa
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue, 20 May 2014 03:48:17 -0400 (EDT)
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <22073306.10715158.1400571893414.JavaMail.zimbra@xxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: Vm6ronsqiGesp9W2M86R0r3CXGleFQ==
Thread-topic: pcp updates: systemd fix, signal fixes, pmchart long options, qa
Changes committed to git://git.performancecopilot.org/pcp.git dev

 GNUmakefile                    |    8 
 build/rpm/fedora.spec          |    2 
 qa/1001                        |    1 
 qa/1002                        |    1 
 qa/1003                        |    1 
 qa/1004                        |    1 
 qa/1005                        |    1 
 qa/1006                        |    1 
 qa/1007                        |    1 
 qa/102.out                     |    4 
 qa/133.out                     |    1 
 qa/304.out                     |    1 
 qa/354.out                     |    2 
 qa/374                         |    6 
 qa/381.out                     |    1 
 qa/538.out                     |    1 
 qa/common.filter               |    2 
 qa/common.test                 |    1 
 qa/group                       |    2 
 src/pmcd/src/pmcd.c            |   56 +++---
 src/pmcd/src/pmcd.h            |    1 
 src/pmchart/main.cpp           |  379 ++++++++++++++---------------------------
 src/pmchart/main.h             |    1 
 src/pmchart/timecontrol.cpp    |    3 
 src/pmdas/systemd/GNUmakefile  |    1 
 src/pmie/pmie_check.sh         |    4 
 src/pmie/rc_pmie               |   10 -
 src/pmie/src/dstruct.c         |    7 
 src/pmie/src/dstruct.h         |    4 
 src/pmie/src/pmie.c            |   24 +-
 src/pmlogger/pmlogger_check.sh |    6 
 src/pmlogger/rc_pmlogger       |    1 
 src/pmlogger/src/logger.h      |    1 
 src/pmlogger/src/pmlogger.c    |   14 -
 src/pmlogger/src/ports.c       |   46 +---
 src/pmproxy/pmproxy.c          |   16 +
 src/pmquery/pmquery.cpp        |   11 -
 37 files changed, 256 insertions(+), 367 deletions(-)

commit 53419bef43c16f13274bb85364d84192631659b0
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 17:42:49 2014 +1000

    Tweak settings in pmquery application for neater output
    
    Two changes here - make the default window size larger for
    the common case of GUI tools generating a usage message;
    and similarly, make the default font used fixed-width else
    we end up with garbled, difficult-to-read output.
    
    Also, when multi-line, scrolled output is produced, scroll
    back up to the start of the text, don't start with with an
    offset at the end - the user will always want to read from
    the start of the message.

commit 2d810bdd597bce18d8b822f9ff38ee8ca145e3ec
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 17:36:49 2014 +1000

    Long option support for pmchart

commit fc44068f8c50514522a00abb3dce8a0290e7d447
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 17:33:47 2014 +1000

    Fix daemon signal handlers to not make use of unsafe functions
    
    It is unsafe to make certain library calls from signal handlers,
    and in several daemons we do this for several different reasons.
    
    Firstly, some daemons trap sigsegv and sigbus to attempt to get
    a stacktrace in their log files.  This should be deferred to any
    system daemons that do this, so we now do that - by default.  In
    certain situations those daemons may not be running or platforms
    may not have such daemons - thus there are situations where we
    may still want to attempt to generate a stackdump in a last-ditch
    attempt at diagnostics.  Thus, these remain as a debug option via
    DBG_TRACE_DESPERATE for those occassions.
    
    Similarly, most daemons printed diagnostics showing which signal#
    they received before exiting, which is also unsafe - again, move
    these to desperate-level diagnostics, off-by-default.
    
    A third case was pmie, which rotated log files on receipt of USR1.
    This is reworked to be deferred to outside of the signal handler.
    
    The fourth and final case was pmlogger handling of volume switches
    via HUP - this has been similarly reworked such that its deferred
    to outside of the signal handler.
    
    This resolves Red Hat BZ #847343.

commit 0dcb2e77cff4319636f2b6de30d2668b5c50c4c2
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 17:24:15 2014 +1000

    Ensure GUI tests source common.config, else xdpinfo check notruns

commit 41ed8d119a30d9a741b8f163c2016239cbda6812
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 17:23:32 2014 +1000

    Handle additional modern distros having /var/run on tmpfs
    
    At some point along the way, Fedora and RHEL began using a non
    persistent filesystem to back /var/run, and so we should not be
    installing files there permanently.  Now handled the same way
    we tackled this on other platforms with similar setups - since
    the scripts can already create the necessary paths themselves,
    on the fly, let 'em do so.
    
    While auditing this, found a couple of cases of needless setting
    of a RUNDIR variable - in the pmlogger and pmie rc scripts.
    
    Also while auditing this, found a case of open coded pcp:pcp use
    in a chmod, instead of using the $PCP_USER:$PCP_GROUP variables
    as all the other scripts do.
    
    This resolves Fedora BZ #656659.

commit 81785d8808739c0108a4162b36c9945e428a4046
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 13:45:34 2014 +1000

    Add X server access notrun checks to each of the pmchart tests

commit 280ef4be6a9c8a44f811d99f0723f1d9448f1f0d
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Tue May 20 13:43:49 2014 +1000

    Mark test qa/233 retired - an IRIX Rebuild script test

commit 5ea3169ca30b27d1fcda583ab7c4f5aa43648e82
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date:   Tue May 20 13:37:00 2014 +1000

    systemd pmda: install $(CMDTARGET)
    
    Commit 8d0b3db3832 in release 3.9.4 removed pmda .so build all right,
    but also accidentally removed the installation of the pmdasystemd
    executable.  Put that back.  The qa/652 test correctly asserts the
    existence of the pmda binary; perhaps leftovers a previous build gave
    a false-positive result.

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates: systemd fix, signal fixes, pmchart long options, qa, Nathan Scott <=