pcp
[Top] [All Lists]

pcp updates: qa, libpcp, mpstat fixes

To: pcp developers <pcp@xxxxxxxxxxx>
Subject: pcp updates: qa, libpcp, mpstat fixes
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Tue, 2 Aug 2016 21:51:17 -0400 (EDT)
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <681807945.13769623.1470189072163.JavaMail.zimbra@xxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: 7D4K6PZAojWcc/Dvkqq7bAWsK9PGGg==
Thread-topic: pcp updates: qa, libpcp, mpstat fixes
Changes committed to git://git.pcp.io/pcp.git master

Ken McDonell (4):
      qa/512: re-audit and re-made .out file
      src/libpcp/secureconnect.c: correct __pmSecureClientHandshake() error case
      qa/admin/check-vm: small changes
      src/libpcp & qa: ...CloseChannel...() Mk II

sitaramshelke (2):
      pcp-mpstat: Fix the machine header to match exact same to the original 
mpstat and fix lowercase 'cpu' in soft_interrupt_usage_reporter
      pcp-mpstat: Update archive with new metrics for printing headers, and 
percpu total interrupts usage as well as update Regression test output

Nathan Scott (1):
      pmlogconf: update mpstat metrics to match those now used


 qa/169.out                                           |    2 
 qa/200.out                                           |   10 
 qa/512.out                                           |   11 
 qa/865.out                                           |    4 
 qa/866.out                                           |   20 -
 qa/883.out                                           |  256 +++++++++----------
 qa/admin/check-vm                                    |   10 
 qa/archives/pcp-mpstat.0                             |binary
 qa/archives/pcp-mpstat.index                         |binary
 qa/archives/pcp-mpstat.meta                          |binary
 src/libpcp/src/connect.c                             |   11 
 src/libpcp/src/context.c                             |   47 +++
 src/libpcp/src/desc.c                                |    2 
 src/libpcp/src/fetch.c                               |    2 
 src/libpcp/src/help.c                                |    2 
 src/libpcp/src/instance.c                            |    6 
 src/libpcp/src/internal.h                            |    3 
 src/libpcp/src/logcontrol.c                          |    8 
 src/libpcp/src/pmns.c                                |    8 
 src/libpcp/src/secureconnect.c                       |   15 -
 src/libpcp/src/secureserver.c                        |   12 
 src/libpcp/src/store.c                               |    2 
 src/pcp/mpstat/pcp-mpstat.py                         |   28 +-
 src/pcp/mpstat/test/interrupt_usage_reporter_test.py |    4 
 src/pmlogconf/tools/mpstat                           |    2 
 src/pmlogconf/tools/mpstat-summary                   |    5 
 26 files changed, 267 insertions(+), 203 deletions(-)


commit 9f3058bd30370e90d582a423d5c21d5775c4050d
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date:   Wed Aug 3 11:48:18 2016 +1000

    pmlogconf: update mpstat metrics to match those now used

commit b16423d79e56f32f15f8e462ee36756a061666c1
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Wed Aug 3 10:01:42 2016 +1000

    src/libpcp & qa: ...CloseChannel...() Mk II
    
    New and improved, even without steak knives.
    
    I've cleaned up the internal APIs ...
    
    - better diagnostics to indicate what was expected as well as what
      was received, namely the wrong PDU (unlikely, now but not so before
      these changes), or PM_ERR_TIMEOUT (most likely) or some other error
      (unlikely)
    
    - renamed __pmCloseChannel() to __pmCloseChannelbyContext() for the
      original use, but also added __pmCloseChannelbyFd() for the second
      round of __pmGetPDU() uses where the context may not be available
      ... this includes the handshake interchanges during context setup
      and pmlc-pmlogger interactions that are built on the same PDU
      services and assume synchronous IPC (but there is no socket
      multiplexing here, so no exposure to the original cascaded
      problems after a __pmGetPDU() timeout.

commit de15fdcf06498d884971f55593c68aafcc8b8510
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Wed Aug 3 09:59:59 2016 +1000

    qa/admin/check-vm: small changes
    
    1. fix certutil line for debian-based systems (broken by some previous
       commit)
    2. drop xclip ... not more than 3 people in 7 billion use the script
       that uses xclip

commit e79b400e02ee4f276db394f6d133085dafcd1ca3
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Wed Aug 3 07:19:24 2016 +1000

    src/libpcp/secureconnect.c: correct __pmSecureClientHandshake() error case
    
    In the case of an IPC, TIMEOUT or PDU protocol botch,
    __pmSecureClientHandshake() was returning -PM_ERR_IPC, not PM_ERR_IPC

commit 28f00a8adc4f4ef3bc0ab95912af51f51923bc55
Author: sitaramshelke <sitaramshelke@xxxxxxxxx>
Date:   Tue Aug 2 23:55:48 2016 +0530

    pcp-mpstat: Update archive with new metrics for printing headers, and 
percpu total interrupts usage as well as update Regression test output

commit fa7f6244a36e94c7bd59c802bcebfcc09fa9fa00
Author: sitaramshelke <sitaramshelke@xxxxxxxxx>
Date:   Tue Aug 2 23:23:38 2016 +0530

    pcp-mpstat: Fix the machine header to match exact same to the original 
mpstat and fix lowercase 'cpu' in soft_interrupt_usage_reporter

commit 3e4f29728aba73bced31b2955bc1ca9bf8fb5eab
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date:   Tue Aug 2 09:02:37 2016 +1000

    qa/512: re-audit and re-made .out file

<Prev in Thread] Current Thread [Next in Thread>
  • pcp updates: qa, libpcp, mpstat fixes, Nathan Scott <=