pcp
[Top] [All Lists]

Re: [pcp] PCP Updates: Bug 1100: pmlogger robustness + New PMAPI Utiliti

To: Dave Brolley <brolley@xxxxxxxxxx>
Subject: Re: [pcp] PCP Updates: Bug 1100: pmlogger robustness + New PMAPI Utilities
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon, 13 Jun 2016 21:04:10 -0400 (EDT)
Cc: PCP Mailing List <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <575B3C4C.9010100@xxxxxxxxxx>
References: <575B3C4C.9010100@xxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: wbrQCljG489QdcjIcaho7+tPY0/5Ew==
Thread-topic: PCP Updates: Bug 1100: pmlogger robustness + New PMAPI Utilities
Hi Dave,

----- Original Message -----
> The new PMAPI calls were created in order to share existing code and are
> similar to existing calls.

Looks good.  As per your suggestion earlier today, I tend to agree this
looks like a 3.11.4 candidate at this stage.

> qa test 326 exercises this code for the case where no PMAPI changes have
> occurred.
> Working on qa for the error scenarios
> 

(see my other note, but I think you are getting close there)

> -----------------------------------------------------------------------------------------------------------------------------
> Changes committed to ssh://git.pcp.io/oss/git/brolley/pcp
> pmlogger-pmda-reconnect
> 
> Dave Brolley (2):
>        New PMAPI calls: pmSemStr(3) and pmSemStr_r(3).
>        pmlogger: Validate all configured metrics after PMCD_ADD_AGENT is
> received     from pmcd.
> 
> src/include/pcp/pmapi.h   |    2
>   src/libpcp/src/exports    |    2
>   src/libpcp/src/util.c     |   50 ++++++++++------
>   src/pmlogger/src/check.c  |  139
> +++++++++++++++++++++++++++++++++++++++++++++-
>   src/pmlogger/src/fetch.c  |   12 +++
>   src/pmlogger/src/logger.h |    4 -
> 

Missing a man page here I think, Dave?  For new PMAPI routines we tend to
have targetted QA tests; if you think the pmSemStr coverage from pmlogger
lacks any (?), something like qa/src/atomstr.c may be the go.

cheers.

--
Nathan

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