On 1.4.0-epa1
I’m not sure whether to classify this as a bug or an Improvement. When configuring a large number of VRFs, the commit time (and checks on reboot) take a long time. This seems to be (at least in part) because the protocols_bgp.py script is called per VRF rather than a single pass for all of the BGP config.
Should I submit this as a bug or improvement for this issue?
It is more improvements
The same behavior is actually for static routes, etc., and for vrf.
Gotcha, thank you! I’ll submit it as an improvement request to have it done in a single pass.
Task created: ⚓ T6122 Protocols under VRF config run in a single pass against their conf_mode scripts
Thanks again @Viacheslav
Probably related ⚓ T5388 Something is fishy with commit and boot times when more than a few hundred static routes are being used
This is a little different. I did some testing for that issue as well and most of that appears to be related to some legacy vyatta functions like creating a revision history. Which if you ever look at is crazy talk. Hopefully that can be removed eventually but who knows how many dependencies there are for that. Hopefully the rollback soft feature is the first step towards that.
This issue will be slightly impacted by that since each VRF will have about 7 lines so it does generate a lot of lines of config. But this has an easier road to improvement since there’s an obvious and actionable course of action in putting all of the actions in a single pass.
Side note, I think it was you that a lot of the digging into that issue; great work. I think it is one of the biggest hurdles towards greater adoption of VyOS.