Hi Jerome,

Comments inline:

On Mon, Oct 08, 2012 at 06:53:09PM +0200, J?r?me Fleury wrote:

> I do have that, and it's working in production. Its limited access to
> our IX members only, but it works quite well, using the isotope.js
> library for the front-end (http://isotope.metafizzy.co)
> I can give you an access so you get the idea.

OK, good. If you can arrange some temporary access to get an idea it
would be great. I assume you did develop yourself the part that takes
counters from RRD and manipulates them so that you can change it to
whatever other backend you might decide to use - correct?

> Sounds like what I'm looking for. But getting a rate instead of a
> counter is also important, because this is the information I need.

OK, point taken for the rate. Meanwhile, here is a snippet to give you
an idea how to aggregate on different temporal resolution - and, by
using dynamic table naming, break down data in different tables, ie.:

acct_5mins_%Y%m%d => 5 mins counters are divided per day in tables
like acct_5mins_20121015, acct_5mins_20121016, etc. Similarly goes
for the other set of tables (see sql_table entry in CONFIG-KEYS doc
for how to expand macros in case you do not find them intuitive).

Note1: MySQL is an example, you could opt for other SQL or noSQL
options; note2: the config is for reference only and is not tested:

..
plugins: mysql[5mins], mysql[hour], mysql[day]
aggregate: <your aggregation method here>
sql_table_version: <depending on aggregate, choose suitable one>
sql_optimize_clauses: true
!
sql_table[5mins]: acct_5mins_%Y%m%d
sql_refresh_time[5mins]: 300
sql_history[5mins]: 5m
sql_history_roundoff[5mins]: m
sql_dont_try_update[5mins]: true
!
sql_table[hour]: acct_hour_%W
sql_refresh_time[hour]: 3600
sql_history[hour]: 1h
sql_history_roundoff[hour]: h
sql_dont_try_update[hour]: true
!
sql_table[day]: acct_day_%Y
sql_refresh_time[day]: 3600
sql_history[day]: 1d
sql_history_roundoff[day]: d
sql_dont_try_update[day]: false
..

Cheers,
Paolo

_______________________________________________
pmacct-discussion mailing list
http://www.pmacct.net/#mailinglists

Reply via email to