|Open||None||T263107 Netbox: Virtual chassis members cables not showing up consistently|
|Open||crusnov||T265084 Upgrade Netbox to 2.10|
I have pushed a new version of the deploy repository with 2.10. Reading the changelog things that will affect our integrations:
- major change to customfields, so anything which interacts with those may need to be changed [they are changed to a json blob with the object they add to]
- there are a number of changes to the API but nothing that immediately stands out, I'll go through them when I go through the tools for testing.
I have deployed 2.10.4 to -next. It worked fine apparently. I'll be doing any necessary porting work but initial tests don't look like there's much. Probably the dump script and perhaps external things like cookbooks need to be poked at (as discussed above).
Of the reports here are some test results:
- cables works
- coherence (errors trying to import CustomFields)
- accounting (errors trying to manipulate custom fields)
- rack (errors trying to import CustomFields)
- managementconsole (errors trying to access connection_status on ConsolePort)
- puppetdb (both work)
None of the cables show their remote endpoints: https://netbox-next.wikimedia.org/dcim/devices/1945/interfaces/
Is there a way to re-generate them?
Okay so it appears to be some weirdness in reproducability for manage.py trace_paths - I was not able to fix it by running this (and rerunning it caused a traceback), but dumping and reloading the database, migrating and rerunning manage.py trace_paths seems to have fixed it. We'll have to make sure this is part of teh production upgrade process so it doesn't get broken in the same way. @ayounsi if you could look around the cables and see if there are any other issues? Thanks.
Also notable is that now that I've done this cleanly, manage.py trace_paths reruns without error. It may be attempting to view the trace without running this does something which breaks future runs or similar. I'll add it to the upgrade process.