How to Deal with Down Devices in Cacti, if Customer does not agreed to Disable or Delete Them ?

[Babai123]

Hi,

How to Deal with Down Devices in Cacti, if Customer does not agreed to Disable or Delete Them ?
 
As Some Devices are manually shutdown by Customer for a specific time period. During that time period, poller impacted due to down devices, and graphs are having cuts.


Any suggestion is helpful.

[cigamit]

Not a terrible lot you can do currently, except to increase the number of threads in the poller, reduce the SNMP/Ping timeouts, and reduce the number of retries.

We have an idea of how to better handle this in the future, where we move the down hosts to its own process so it doesn't lock up the rest of the threads.

[Babai123]

Hi,

Now we are having maximum number of process and threads.
8 processes & 50 Threads per each process.
SNMP timeout is 400 for all Device, if we will decrease the time out. we may miss data from Device.

Any Suggestion is helpful for us.

[Osiris]

I'm just curious why the customer wants to preserve them if they are gone? If they are gone should they not then be disabled?

[Babai123]

Hi,

Some Customer Switch of the Devices for 2 -3 months as per their requirements or Holidays . So they don't want to disable or Delete them. Any Solution for these type of Scenarios ?

Any suggestion is helpful for us.

[cigamit]

Seems odd to let your customers dictate your own internal business processes. I already gave several suggestions. The only other thing I would think of doing is setting up a "remote" poller on a separate instance and just assign that customers devices to it (or assign every down device to it via a script / sql query or write a custom plugin).

[TheWitness]

Any rationale human being would "disable" hosts that are no longer present, or will no longer be available. After that, you can still view legacy Graphs. There is an option to hide graphs from disabled devices, but it's a user level setting.