Interface bandwidth percentage on downed cisco interfaces

[xnrl]

Hello,

I'm using the Threshold plugin to calculate bandwidth utilization (via CDEF + ifHighSpeed) and have recently run into a problem with downed interfaces on Cisco commutators.

The value of downed interface speed they send via SNMP is always 10Mbit/s. So I assume this is what happens: the inactive interface speed is cached by Cacti, THold calculates current bandwidth as 0, then the interface comes up, cached value is still the same -> percentage goes through the roof with like hundreds %. Verbose query on the device / query refresh / cache rebuild fixes the problem for a while, but of course it pops after some time. And even considering the number of Cisco devices monitored is only around several dozens, receiving several false positives a day and going to cacti just to fix them is not very refreshing.

My only idea on how to fix this permanently is to modify THold code to call rebuild_poller_cache.php --host-id <host id> after calculating a value above 100, then repeating the calculation. I'm not very eager to do this though 'cause it's bound to be forgotten/broken when we next decide to update THold.

Maybe there's some way to fix this with standard THold / Cacti capabilities? I'd be grateful for any suggestions.

CDEF for bandwidth %: cdef=CURRENT_DATA_SOURCE,8,*,|query_IfHighSpeed|,1048576,*,/,100,*
Using Cacti 0.8.8a with cacti-spine 0.8.8a-1 and THold 0.5.

[Happiness0]

I have 2 suggestions:

- Call the 'rebuild_poller_cache.php --host-id <host id>' command with a scheduler (cron on Linux for example)

- Do not rely on CDEF. Use another tool to calculate the percentage, like a custom data query or data input method