Page MenuHomePhabricator

Netbox: investigate custom status
Closed, ResolvedPublic

Description

The upgrade to Netbox 3.2 brings Customs status:
https://docs.netbox.dev/en/stable/configuration/optional-settings/#field_choices

Until now we've been bound to use the set list of Netbox status: https://wikitech.wikimedia.org/wiki/Server_Lifecycle#States but it's now possible to fully customize them, for any models (objects), and similar to tags it can be a slippery slope.

I see 3 options for servers:

  • Don't change anything, as we already have our tooling and habits set on the default status
  • Keep the current status and add additional ones
  • Rename the existing ones to match our in house status

The last 2 options would need matching changes in our doc and automation.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript

I would not rename them and keep the netbox upstream list as a base, and optionally add 1~2 states if we see fit. For example a reimaging/PXE/maintenance state could be a good candidate (TBD a good name).

@wiki_willy please let us know if DCops have any preference on this topic.

Thanks for looping me in @ayounsi, I'll check with the team and circle back afterwards.

@wiki_willy please let us know if DCops have any preference on this topic.

I think the naming convention of the current states are fine the way they are. In terms of adding additional states, we were wondering if it makes sense to create a Netbox state called "recycled" for moving recycled hardware to this state, instead of deleting them from Netbox? (ref: T320955) Also, there have been some initial conversations with Deb Tankersley (through Tim Starling) about possibly targeting and powering off any long standing idle servers in an effort to reduce power consumption. If this were something doable, we could potentially need a state called "idle" or something to distinguish these servers from other production servers. @Volans - I'm not sure if Joanna has reached out to you about it yet, but that's a bigger conversation that could tie into this. Thanks, Willy

ayounsi claimed this task.

Thanks! I replied on T320955 about the RECYCLED status.

The IDLE one could be an option to discuss if the need arises (now we know we have this tool in our toolbox).

For the scope of this task I think we can close it. And follow up in T320955 or a new one if needed.

Also, there have been some initial conversations with Deb Tankersley (through Tim Starling) about possibly targeting and powering off any long standing idle servers in an effort to reduce power consumption. If this were something doable, we could potentially need a state called "idle" or something to distinguish these servers from other production servers. @Volans - I'm not sure if Joanna has reached out to you about it yet, but that's a bigger conversation that could tie into this.

@wiki_willy yes Joanna has mentioned it to me, although I'm a bit skeptical as this "idle" state should just not exists and hosts that are provisioned and handed over to a service owners should just be put into production in a timely manner and powering them off would have quite some drawbacks. That probably deserve a separate task though.