pcp
[Top] [All Lists]

Re: nginx PMDA error

To: Tadej JaneÅ <tadej.j@xxxxxx>
Subject: Re: nginx PMDA error
From: "Frank Ch. Eigler" <fche@xxxxxxxxxx>
Date: Wed, 9 Sep 2015 11:35:47 -0400
Cc: pcp developers <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1441811714.27574.80.camel@tlinux64>
References: <1441613884.8358.33.camel@tlinux64> <y0m37yqkqph.fsf@xxxxxxxx> <1441790097.27574.20.camel@tlinux64> <20150909125102.GD16438@xxxxxxxxxx> <1441811714.27574.80.camel@tlinux64>
User-agent: Mutt/1.4.2.2i
Hi -

> [...]
> Here are the contents from /var/lib/pcp/pmcd/nginx.log:
> Log for pmdanginx on collector started Wed Sep  9 15:07:57 2015
> 
> [Wed Sep  9 15:07:59] pmdanginx(25446) Error: pmdaFetch: PMID 117.0.6
> not handled by fetch callback
> [...]

I'm really not familiar with the insides of that pmda, but there
is not that much going on.  Check it out - it's just a short perl
script.  It does a http fetch on "http://localhost/nginx_status";
(or some other location overridden in eval'd nginx.conf).  The
resulting data is parsed with a blunt

     @nginx_status = ($response->decoded_content =~ m/(\d+)/gm);

and then metrics are fetched from that array, indexed by
the item number (e.g. 6).  

So if I had to diagnose this bad boy, I'd start with a hand-fetched
nginx_status result file, and insert some tracing to dump
@nginx_status after the parse process.


- FChE

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