pcp
[Top] [All Lists]

Re: [pcp] json pmda qa/1052 failing on Centos6.6

To: Nathan Scott <nathans@xxxxxxxxxx>, David Smith <dsmith@xxxxxxxxxx>
Subject: Re: [pcp] json pmda qa/1052 failing on Centos6.6
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon, 10 Aug 2015 16:13:03 +1000
Cc: PCP <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <487433945.5357261.1438845771574.JavaMail.zimbra@xxxxxxxxxx>
References: <017301d0cc28$d7c78640$875692c0$@internode.on.net> <440072288.2807362.1438581950952.JavaMail.zimbra@xxxxxxxxxx> <55BFC26D.8040503@xxxxxxxxxx> <1889100767.3736026.1438652097204.JavaMail.zimbra@xxxxxxxxxx> <55C14FAF.9010100@xxxxxxxxxxxxxxxx> <934995702.4411085.1438732642013.JavaMail.zimbra@xxxxxxxxxx> <55C15F12.8060408@xxxxxxxxxxxxxxxx> <1837540003.4468854.1438736913075.JavaMail.zimbra@xxxxxxxxxx> <487433945.5357261.1438845771574.JavaMail.zimbra@xxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.8.0
On 06/08/15 17:22, Nathan Scott wrote:
..
Looking again, there is a check in configure.ac that attempts to disable the
pmdajson build if the python version is <=2.6 - it seems like that should've
protected us here...

Oh - I wonder if its the makefile ignoring this?  Can you check output from
rpm -qf /var/lib/pcp/pmda/json there Ken?  If its "pcp" & not "pcp-pmda-json"
then this all makes alot more sense, and we just need to tweak the makefile
for src/pmdas/json such that it honours PMDA_JSON from builddefs.

Afraid not ...

kenj@vm14:~$ rpm -qf /var/lib/pcp/pmdas/json
pcp-pmda-json-3.10.5-19.x86_64

Ahhh ... but this is 3.10.5 ... consider this

kenj@vm14:~$ rpm -qf /var/lib/pcp/pmdas/sample
pcp-libs-devel-3.10.7-2.x86_64

kenj@vm14:~$ grep JSON /usr/include/pcp/builddefs
HAVE_PYTHON_JSONPOINTER = true
PMDA_JSON = false

Looks like I just need to explicitly remove the old pcp-pmda-json ... I wonder why this did not cause all sorts of rpm compaints, surely there are some dependencies that are needed, but perhaps they are >= 3.10.5 and so the new packages satisfy the dependencies?

Any way, real punters would probably not have seen this issue.

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