I’m currently working on integrating Cambium Access Points into our Zabbix monitoring setup. Specifically, I’m looking to monitor throughput (both uplink and downlink) on these devices. However, I’m having difficulty locating the correct OIDs to accurately track this data.
I’ve been working with the CAMBIUM-ePMP-5.8.0-RC27-MIB file, but I haven’t been able to identify the precise OIDs that correspond to throughput metrics. If anyone has successfully monitored Cambium AP throughput in Zabbix, I would greatly appreciate it if you could share the relevant OIDs or, even better, a template that includes these items.
Additionally, any tips or insights on best practices for monitoring Cambium devices in Zabbix would be invaluable.
I am fairly certain you will not find an OID for throughput and need to look into the math behind the common ifInOctets and ifOutOctets and tracking timestamps for when those OIDs are gathered which you will likely find for every interface in any device you are looking to monitor.
IF-MIB::ifHCOutOctets = .1.3.6.1.2.1.31.1.1.1.10
and
IF-MIB::ifHCInOctets = .1.3.6.1.2.1.31.1.1.1.6
which return a 64 bit count of number of bytes since interface reset, like so:
IF-MIB::ifHCInOctets.1 = Counter64: 0
IF-MIB::ifHCInOctets.2 = Counter64: 20148104652
IF-MIB::ifHCInOctets.3 = Counter64: 2563504123
IF-MIB::ifHCInOctets.4 = Counter64: 2563504123
That gets recorded into RRDtool along with the timestamp of when the data was gathered which on data read gets measured as a rate in any speed unit we desire.
RRDtool is an option. I think RRDtool data can be imported into Zabbix. I have tested Zabbix a few times but am not an expert in using it. I think it would likely have its processes to do many of the things RRDtool does in handling math calculations and situations like user counter resets and reboots. But, I don’t think Zabbix easily integrates into displaying data collected from RRDtool as it is running but I might be wrong.
Another thing to keep in mind about throughput is that because it is calculated by comparing timestamps of collected data at certain times each reading represents an average. And, a lot of traffic would demonstrate, if analyzed more carefully with higher granularity, higher peaks and troughs throughout the same period with more readings in that period. Tools like RRDtool, if not RRDtool, also often archive data, and the graphs as time goes on will show less of the peaks and troughs and often just keep and display the averages for longer periods.
I supposed if we wanted to we could define something like a Chicken Nugget per milliLustrum instead of kibibytes per second where a Chicken Nugget would be 1,024 times 1,000 bytes. I am from CA. Keep in mind a lot of the confusion regarding how a kilobyte can somehow be 1,000 bytes when pretty much every computer science program, I assume, makes it abundantly clear it is based on binary and is 1,024 bytes because the CA legal system wouldn’t do their duty and assert a measurement of binary data must be base 2 and not base 10 and allowed clear false advertising that has crept into measurements for data transfer speeds. However, I’m pretty sure agencies like the FCC still ask for bits per second not necessarily accepting CA legal nonsense, at least IMO. While this forum may not be the place for this discussion it would not surprise me at all if when CA allowed large corporations to falsely advertise data storage amounts the math proficiency of CA students on average also rapidly declined after that.
I wonder how many CNpmL (Chicken Nuggets per milliLustrum) there are in a megabyte per second if we accept 5 years as 5 times 364 1/4 days?