Alternate reporting

[Patrice]

Hi,

First of all thank you very much for making Jiffy open source. We're
now using it on the http://www.local.ch/ platform since Monday.

As part of that implementation I decided to not use any database and
instead directly plot the relevant values using MRTG. For that I
implemented a script which takes the log lines on standard input and
every five minutes writes out the current numbers to the file system.
The script is configured in Apache as the log file (using Apache piped
logs - http://httpd.apache.org/docs/2.2/logs.html#piped)
The script is at <http://paste.pocoo.org/show/78432/>. The whole
parseLogEntry() function has been ported from the Perl script you
provide.

MRTG then uses a script to read the data from that file and plots the
values. The result then looks like this: http://www.flickr.com/photos/patrice/2632812489/

I hope this script is useful to somebody else as well.

Cheers
Patrice

[Scott Ruthfield]

Hi Patrice -

Very glad to hear that you're using Jiffy! It was nice to see it in your production code. :)

This is a nice simplification if you don't want to deal with databases, and the code seems easily adaptable if you want to store different reportable metrics off of the raw data (median, 90th percentile, etc.). While our reporting tools didn't provide anything besides average (yet), in my experience there are other core data points that are at least as important to see quickly.

This does seem to assume that you're using just one web server, if I'm reading it correctly?

-Scott

(P.S. I've heard of some other interesting mods coming...)

[Patrice]

> This does seem to assume that you're using just one web server, if I'm
> reading it correctly?

It does work with multiple servers. MRTG collects the data from all
our frontend servers and calculates the correct average over the whole
cluster.

Cheers
Patrice