Graph over all discovered values

[toomask]

Hi.

Lets say i have discovery item which measures in bytes and has no limit or upper value (like disk with unknown size). This item dicovers number of instances and i want pie graph which represents relative weight over all produced numbers. Since number of instances is not known before actual discovery and they can change over time, i must define pie in discovery template. Is it possible?

Actual device is Data Domain 670 and monitored item is mtree size before compression, but explaining that terminology is out of scope. I can upload template later, once i get everything working.

[steveboyson]

You cannot automatically graph distinct items found in separate discovery rules within on graph or pie. Simply not possible.

You can do it by hand on the host level after all items have been discovered or use the API after successfull discovery to create the pie graphs.

[toomask]

items are NOT in distinct discovery rules. These are instances of the same item in the same discovery rule.

It is like discovery rule for SNMP DATA-DOMAIN-MIB::mtreeCompressionMtreePath which discovers number of paths (or discovery item instance) with #SNMPINDEX 0..20+. There is one discovery item in it: SNMP DATA-DOMAIN-MIB::mtreeCompressionPreCompGib.{#SNMPINDEX}, which produces numeric value for each path.

Question is still the same -- how do i combine these values into one pie graph at template level?

[steveboyson]

items are NOT in distinct discovery rules. These are instances of the same item in the same discovery rule.

It is like discovery rule for SNMP DATA-DOMAIN-MIB::mtreeCompressionMtreePath which discovers number of paths (or discovery item instance) with #SNMPINDEX 0..20+. There is one discovery item in it: SNMP DATA-DOMAIN-MIB::mtreeCompressionPreCompGib.{#SNMPINDEX}, which produces numeric value for each path.

Question is still the same -- how do i combine these values into one pie graph at template level?

You did not read carefully. An item prototype is instanciated and becomes a 1..n distinct items depending on the number of found values in the discovery rule.
At the time when the discovery rule is running it is unknown how many items will be found, thus you cannot create a graph consisting of all found items.

Again: I think this is not possible within a template or a LLD definition.

[toomask]

Thank you for explanation. I suspected myself too, that at item prototype level there is no knowledge about other (possibly not yet) discovered items.