In tor-scale setup, a vn's RI is not set correctly in control-node
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R2.20 |
Won't Fix
|
Medium
|
Sachin Bansal | |||
R3.0 |
Won't Fix
|
Medium
|
Sachin Bansal | |||
R3.1 |
Won't Fix
|
Medium
|
Sachin Bansal | |||
Trunk |
Incomplete
|
Medium
|
Sachin Bansal |
Bug Description
R2.1 Build 44 Ubuntu 14.04 Multi-node icehouse setup
nodei38 has ~140 Tor-agents.
vn1 has ~11k vmis and 1.1K BMSs are sprayed across 110 TORs(ovswitch vms) in the same VN
vn129 to vn162 has 1K VMIs each (Each TOR having 1K vmis in separate VNs)
i.e.
vn129 is configured on TOR ovs-vm129 with 1K vmis (10 ports) and so on
While contrail-api was getting initialized, i had restarted supervisor-config on nodei34.
It was seen later in control-node that RI of vn1 has deleted state set to true and peering set to Tor-agent "nodei38-147" which should have never happened.
The right ifmap config for vn1 was not downloaded to agent at all.
Tor-agent nodei38-147 should never have got config for vn1
-------------------
The ifmap config of this tor-agent shows only this :
type:routing-
You can check http://
-------------------
supervisor-config is stopped on nodei35 for ease of use
env.roledefs = {
'all': [host2, host3, host4, host5, host6],
'cfgm': [host2, host3],
'openstack': [host2],
'webui': [host3],
'control': [host3, host4],
'compute': [host5, host6],
'collector': [host2, host3],
'database': [host2, host3, host4],
'toragent': [host6],
'tsn': [host6],
'build': [host_build],
}
env.hostnames = {
'all': ['nodei34', 'nodei35', 'nodei36', 'nodei37', 'nodei38']
http://
tags: |
added: config removed: vrouter |
Reduced the priority to "medium" as we have seen this only once.. -- Ashish