SNMP: APs not using a distinct sysObjectID for each model

  • 1
  • Question
  • Updated 1 year ago
Is anyone using a 3rd-party NMS with Aerohive AP's? I've been looking at the SNMP implementation of the AP products and whas surprised by the fact that SNMPv2-MIB::sysObjectID value returned by all AP models is the same, while it's best practice (and to return a value that maps to specific models. Generally the sysObjectID is also used by NMS tools to determine what kind of device they're dealing with and for inventory purposes.

What puzzles me is that the structure of AH-SMI-MIB was actually designed to hold specific model numbers; there's a hierarchy for Aerohive products (ahProduct), Aerohive APs (ahAP) and then 4 models are defined, namely AP020, AP028, AP320 and AP340. However all HiveAP's that I've seen in the field report the generic "ahProduct" OID.

I wonder why Aerohive decided to create a proper structure to record models in the AH-SMI-MIB, but then didn't follow through with adding newer AP's to the MIB and why it was decided to return a generic value for sysObjectID.

Full disclosure: I'm a software developer, focused on multivendor network management and security; contributor to open source NMS tools Netdisco & SNMP::Info.
Photo of Jeroen van Ingen

Jeroen van Ingen

  • 4 Posts
  • 0 Reply Likes
  • confused

Posted 3 years ago

  • 1
Photo of Nick Lowe

Nick Lowe, Official Rep

  • 2491 Posts
  • 451 Reply Likes
An inadvertent oversight probably, hopefully it can be fixed for 6.5r3 and 6.6r2.

It would need to be mentioned in the release notes though as there could be compatibility issues with existing deployments as people update. It ought to be done though.

Nick
Photo of Mike Kouri

Mike Kouri, Official Rep

  • 1030 Posts
  • 271 Reply Likes
Why is our MIB messed up? Because we had some developers in the past who didn't really understand what they were doing, and then because no one complained enough to justify the effort to re-do things. 

Nick is correct that we do intend to address this in future releases of HiveOS, but he is overly optimistic about when we'll be able to get that cleaned up (sysObjectID is not the only goof). 

We will probably have to wait for a major break in version numbers (i.e. jump from 6.x to 7.x) before introducing such substantial changes, so don't expect changes in this area this calendar year.
Photo of Jeroen van Ingen

Jeroen van Ingen

  • 4 Posts
  • 0 Reply Likes
Thanks Mike! While reporting a more specific sysObjectID isn't a major change in itself, I can imagine that it may have unforeseen consequences in "surrounding" tooling. So yeah, sounds logical to wait for a new "major" version number like 7.x.

I sincerely hope that this will be fixed in such a new major release.

Given the historic release numbering etc, am I right to assume that now the 6.x major branch is already up to 6.5, Aerohive is nearing the end of the 6.x life cycle and a 7.0 is already planned or maybe even being actively developed?

Or, more to the point: what are the chances that this will be addressed in the next calendar year?
Photo of Jeroen van Ingen

Jeroen van Ingen

  • 4 Posts
  • 0 Reply Likes
I heard a rumour that specific OIDs for each device are forthcoming. Is it possible to get the list in advance, to prepare my NMS?
Photo of Mike Kouri

Mike Kouri, Official Rep

  • 1030 Posts
  • 271 Reply Likes
I am horribly sad to report that the rumors were worng. 

We did have intentions to finally address this in the upcoming releases of HiveOS, we even began the work, but due to some personnel issues that came up, that had to be deferred to a future release. 

Please accept my assurances that I really REALLY want to get this done, as soon as possible.