Granularity of history returned via API

[jaragunde]

Hi,

I'm building a custom interface on top of Zabbix API, and I'd like to retrieve history data to build custom graphics.

The problem is that if I'm showing a big interval of time, I get too many values (for example, retrieve a year of hourly-sampled data) and eventually make Zabbix run out of memory.

Is there a way to specify the granularity of the data? Following the previous example, gather a year of data taking only one value per day.

[nelsonab]

http://www.zabbix.com/documentation/1.8/api/history/get

There are two parameters to try, time_from and time_to. They both take Unix time as their parameters.

Hopefully that helps.

[jaragunde]

I was thinking on something different, for example a way to specify a "step", and retrieve only one value on every ten stored values, for example.

I guess that Zabbix web interface also has to do something similar when it plots a graph with a big zoom frame: not every value is shown, instead they are grouped and an average of every group is shown.

[jaragunde]

It's an important feature for me, so I've started working on a patch. I'm taking inspiration from frontend/php/include/classes/class.cchart.php, function selectData(), which is used to gather history data to build charts in the web frontend.

I've created a ticket: https://support.zabbix.com/browse/ZBXNEXT-656

[r-br]

Such a feature is interesting to me as well.