Netbox: Cable traces do not display `via` anymore

[hellerve]

Environment

• Python version: 3.7 (should be on any)
• NetBox version: 2.8

Steps to Reproduce

1. Disable any installed plugins by commenting out the PLUGINS setting in
   configuration.py.
2. Add two devices with interfaces, let’s call them A and B.
3. Create a circuit with two terminations, one on the A side and one on the Z side.
4. Connect interfaces on device A to the A side of the circuit, and device B on the Z side.

Expected Behavior

Historically, the circuit termination was the connected endpoint, meaning that you were able to see the property via.

Here’s an interface connected before the update:

This is the trace:

Observed Behavior

Due to the new tracing logic, circuit terminations gets resolved in there, meaning that the template never ends up in the code path that renders the via, and only interfaces are the connected endpoints.

Here’s an interface connected after the update:

Here’s the trace:

This behavior change is a bit tricky to resolve, since it involves two parts of the Netbox code: the signal that handles cables and connected endpoints in netbox.dcim.signals:update_connected_endpoints, which after a bit of indirection calls trace, and the template netbox/templates/dcim/inc/interface, specifically the part that triggers on iface.connected_endpoint.term_side.

I hope that this is understandable. I’d be happy to provide any additional info that you might need.

Cheers

[jeremystretch]

Create a cable from an interface to another, through a circuit.

Please provide the exact steps you are taking to reproduce this issue.

[hellerve]

I’ve updated the description to be a little less terse.

[DanSheps]

Looks like #4622 and #4625 might be related.

[DanSheps]

Doing some testing, seems like the problem is straight forward:

When the path is first initialized, the path doesn't have two endpoints. Once the path is fully initialized (two terminations connected), it only cares about the end path.

I think this is going to be a weird corner case where we are going to need to also check along the path and cache the the appropriate sides for a circuit termination in the path the endpoint as well. Not sure how easy the fix will be though, as you also need to take into account circuits that are in-between when deleting or updating the path.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. NetBox is governed by a small group of core maintainers which means not all opened issues may receive direct feedback. Please see our contributing guide.

[steffann]

I'm still working on this, but it's harder than I thought so I need a bit more time.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. NetBox is governed by a small group of core maintainers which means not all opened issues may receive direct feedback. Please see our contributing guide.

[steffann]

Silly stalebot. I'm working on this. Honest!

A little more detail: I am testing with storing trace results so they are easy to display. First results look promising.

[martink2]

@steffann, Thanks for trying to get this sorted out. We do very much rely on the circuit visibility
on the device interface list. We currently have held off adding a large number of circuits which have been recently installed pending the outcome of this fix. If we would add the circuits now would your fix show them correctly on the interfaces after the update or is this only being evaluated when we disconnect/re-connect the circuits later?

[steffann]

My intention is to require no manual action

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. NetBox is governed by a small group of core maintainers which means not all opened issues may receive direct feedback. Please see our contributing guide.

[hellerve]

Could you use any help with anything, @steffann? Unfortunately I’ll be on vacation next week, but I can help starting the week after that :)

[Kircheneer]

@hellerve It looks like #4706 should resolve this as @DanSheps mentioned there.

[hellerve]

@Kircheneer Ah, thanks! I seem to have missed that backreference here.

[tyler-8]

I did some testing on #4706 (now merged) and it does not appear to fully fix this particular issue.

I first tried editing the cable and the circuit terminations and click the 'Save' button with no change. I also disconnected the cables from either termination and re-created them with no change.

However, when I deleted the circuit terminations and re-created them, as well as recreated the interface connections, I got the termination A on the circuit detail page to show the device and interface it was connected to. The termination Z did not show the device/interface data.

And on each device's detail page, the "via" was still missing, only showing the distant end device name and interface as before.

For clarification here's the path I'm working with:

+---------+     +-----+     +------------+     +------------+     +-----+     +---------+
|Interface|-----|Cable|-----|CircuitTermA|-----|CircuitTermZ|-----|Cable|-----|Interface|
+---------+     +-----+     +------------+     +------------+     +-----+     +---------+

[ddragic]

Haven't done any extensive testing, but I just upgraded to v2.8.7 and as @tyler-8 said this issue is still there.

[steffann]

It turned out that with the current database models it is not feasible anymore to show "via" for interface-to-interface paths. What is still possible is to show the circuit when the trace ends at CircuitTermA (connected endpoint shows the circuti) or CircuitTermZ (connected endpoint shows "site Z via circuit X".

This is part of a recent cleanup ticket. I'm sorry that we cannot restore the "via" functionality for this use case at this point.