1.1.3 graph y axis calculation

[markpr]

Hi,

Just getting ready to move to the latest 1.1.3 (from an old beta) and this release is looking really good. Thanks Alex and Team.

I'm already pulling good data (values) from a simple check using the icmppingsec function but the y axis scale isn't displaying correctly.

Looking at Monitoring | Latest Data | and the raw data in "Values of a Specified Period" I get ping values of 0.2550 0.2570 etc etc. If I graph this by selecting "Graph of a Specified Period" the graph looks good apart from the Y axis which has a scale of 00:00:00.

I get the same thing if I try and create a simple graph from this data whether I set the Y axis to be calculated or fixed. If I select fixed and set the minimum and max values then I can see that this does influence the Y axis because the scale seems to change to 00:00:01 and 00:00:02. (assume this is HH:MM:SS).

These are all very basic graphs and I have these (and more complex graphs) working fine in 1.1beta6 but this has me stumped.

Any ideas? (I have attached two graphs to show the two types below). I'm running the latest apache 2.2.3 with PHP 5.1.6 and MySQL. Nothing in the apache, zabbix, or php logs indicate an error.

I couldn't find anything else in the forums but have applied the patch for 1.1.3 functions.c in the forum sticky (this did not help this problem)

thanks

Mark

[markpr]

I have either resolved or worked around the problem by the following:

for the my icmppingsec() item definition:

used a customer multiplier of 1000
changed the units (label) to ms

I'm now storing values in milliseconds (e.g. 250ms) and the axis seems to set the correct scale although the scale labels are odd in how they intermittently display precision (e.g. 80.00ms 160.00ms 240ms 320.00ms 400ms 480ms)

This works for me but is this a bug in how the graph calculates scale for floats?

Mark

[Alexei]

It should be fixed.

[markpr]

Alex

Thanks !

BTW The workaround also caused a problem with the associated trigger so I had to update this temporarily as follows:

{HOSTNAME}:icmppingsec.last(0)}>1000 (milliseconds)
instead of
{HOSTNAME}"icmppingsec.last(0)}>1 (second)

otherwise the trigger which is intended to alert when ping latency is greater than 1s will be constantly TRUE.

Mark