Home

Awesome

Termination of Project

Over the recent years, cacti has matured and has absorbed this idea into a new "Bulk Walk Maximum Repetitions" feature, which offers an auto detect method to determine the optimal number of repetitions.

While this is sufficient for most uses, there are some cases such as F5 load balancers, where manual intervention is still required in order to not exceed payload maximums.

As such, i will be terminating this project. It was fun. Check out a still-very-useful spinoff project called "OIDrage" at https://github.com/patrickscottbest/OIDrage

Thanks!

Patrick 20221221.

Download

https://github.com/patrickscottbest/hammerOID/archive/hammerOID.zip

Updates

20221208 - I intend to pull this project up from beyond the grave. 20221221 - Abandoned.

About hammerOID

hammerOID is a Cacti plugin designed to give insight to overall poller health and provide automated machine learning performance tools for setting optimum Maximum SNMP Object Identifiers (Max OIDs).

  1. Insight is provided defacto with graphs based on internal table information provided by Spine poller. This by itself is a very useful feature of hammerOID plugin.
  2. Automatic OID "hammering" is not turned on by default, but provides some amazing "overnight" insight into the performance of end equipment when OIDs are shifted.

The problem with OID selection in Cacti is that it is often overlooked and depending on the scenario or end-host equipment either contributes very greatly or insignificantly to the overall polling health of the cacti ecosystem.

In large deployments (>100,000 data sources) or in installations where critical polling interval is too close for comfort (just slightly less than 5 minutes), this script can greatly reduce the overall time required to compete a poll cycle.

Challenges encountered with OIDs:

*It is most often left to defaults, or is adjusted without knowing exactly what is happening behind the scenes. *Some devices lock up (IBM AMM cards for example) if the requested OIDs are too low *Some devices use exorbinantly large OID strings which cause rework of the Spine poller to make requests that fit in a single UDP query (F5 VIP OIDs are very lengthy) *A sweet spot exists to reduce overall poll time on a per-host basis, which can be found through experimentation, but that's time consuming

A previous article I wrote on how to tune OIDs and the theory behind it. http://realworldnumbers.com/cacti-tuning-how-to-set-maximum-oids-per-get-request/

Download

https://github.com/patrickscottbest/hammerOID/archive/hammerOID.zip

Current Release

Release Notes

Purpose

Features

Prerequisites

Installation

Notes

Procedure:

Uninstall

Usage

Additional Help?

Possible Bugs?

Copyright

Copyright 2017 Patrick Scott Best

Contact

patrickscottbest@gmail.com