Interface vrf_name Error and are inactive for ~15min

[Vijay Anand]

Contrail Version : 2.21 (14)

Problem:
======
tap interface state is inactive and vrf_name is "Error" for ~ 15 min.

Condition:
========
Setup was UP and running with 200+ SI (i.e. 200+ Heat stacks). Deleted each stack at 1 sec interval, all stacks are cleaned.
After couple of minutes
Launched 30 instances in burst (from CSO application)

Observation:
==========
CSO (VNF Manager) reported all SI are unreachable and initiates rollback (stack delete) for all SIs.

At Contrail during problem state,
- SIs are launched successfully
- No IP assigned to the vSRX(SI) interfaces enabled with dhcp-client
- Above issue is seen for all SIs (30)
- Manually assigned IP to one instance and tried pinging gw - didn't work
- Noticed tap interface is inactive and the vrf_name "Error" for all SIs

at vSRX, dhcp times out and stops sending requests..

After ~15 min, noticed tap interface are active and got VRF assigned

Impact:
=====
Though Contrail recovered after few minutes.. this impacts CSO and the SI.