pcp
[Top] [All Lists]

Re: [pcp] pcp updates: early registration of anon metrics for event reco

To: Nathan Scott <nathans@xxxxxxxxxx>
Subject: Re: [pcp] pcp updates: early registration of anon metrics for event records
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Wed, 13 Jan 2016 10:49:20 +1100
Cc: pcp@xxxxxxxxxxx
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1630098058.7796995.1452567225360.JavaMail.zimbra@xxxxxxxxxx>
References: <56940002.2020004@xxxxxxxxxxxxxxxx> <1630098058.7796995.1452567225360.JavaMail.zimbra@xxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.4.0
On 12/01/16 13:53, Nathan Scott wrote:


----- Original Message -----
Addresses http://oss.sgi.com/bugzilla/show_bug.cgi?id=1133

Changes committed to git://git.pcp.io/kenj/pcp master


Looks good to me.  I'm seeing a little bit of QA fallout ...

720.out vs 720.out.bad
101,102d100
< event.flags
< event.missed

823.out vs 823.out.bad
10c10
< sample.control -21357 Authentication - authentication failure
---
sample.control -12357 Unknown metric name

956.out vs 956.out.bad
20c20
< hinv.ncpu -12389 Try again. Information not currently available
---
hinv.ncpu -12357 Unknown metric name


Looks fairly minor though.

Thanks Nathan.

720 - I've fixed this one ... it is related to the code changes described below for 823.

823 - this is hiding a side-effect of the anon registration changes that I want to change ... there is no need for the pmcd roundtrip to check (and fail as expected) the name lookup of the anonymous metrics.

I have a code change for this and 823 passes now for me.

956 - I have a lot of trouble with this one, I cannot make it stable ... but the code change described for 823 reverts the pass/failure signature what I've been seeing on and off for weeks (at least) now.

I have commits ready to go and tested with check -g pmns -g derive ... just want a couple of full QA runs to make sure there is no other fallout.

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