Detecting gap in graph

[wizou]

Hi Cacti community ,

I am using Cacti to graph several servers in my network and I need to develop a script which can report each time datas cannot be collected (resulting in a gap in the graph).

I have found two clues to detect these gaps :
> From the log file
> From the database directly

For the log file, i have put the logging level to debug/high and i haven't found anything consequent to determine if a gap is occuring.

For the database, i have found that if datas haven't been retrieved, the following query : "select * from poller_reindex where assert_value = 'U';" will show the different host_id where there was a problem. The value "U" means that the last value recovered is Unknown.

But still, i'm not sure if the result of my search are reliable.

So my question is : Do you know if there is a simple and reliable way to determine whenever Cacti cannot retrieve data from a specific host ?

Thank you in advance.

[TheWitness]

What version of Cacti?

[wizou]

I am using the version 0.8.7b of CACTI.

[TheWitness]

Keep logging to LOW, and look for WARNING's in the Cacti Log.

TheWitness

[wizou]

I have done a test which consists to desactivate the SNMP service of a graphed server.
When the logging level is set to LOW, the log doesn't show any warning nor error.
However the graph of this server clearly show a gap.

I investigated in the log and the database, but the only place I have found which tell there is a gap is from the table poller_reindex.

Is this normal ? or can this phenomenon be caused by the "Downed Device Detection" option which is set to "Ping and SNMP" ?

Is there any other way to tell when there is a gap in the graph ?

[TheWitness]

Set avail to SNMP

[wizou]

Thank you for the answers

I lowered the "failure count" and the "recovery count" to 1 and i found what i need in the log.