[CentOS 1.20-55~icehouse.el6] Unable to ping remote VM via a multi-stage Service Chain with ECMP
Bug #1381978 reported by
Ganesha HV
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R1.1 |
Fix Committed
|
High
|
Naveen N | |||
Trunk |
Fix Committed
|
High
|
Naveen N |
Bug Description
Setup:
=====
env.roledefs = {
'all': [host1, host2, host3, host4, host5],
'cfgm': [host1],
'control': [host2, host3],
'compute': [host4, host5],
'collector': [host1],
'openstack': [host1],
'webui': [host1],
'database': [host1],
'build': [host_build],
}
env.hostnames = {
'all': ['nodec4', 'nodec5', 'nodec26', 'nodei27', 'nodei28']
}
1]. Created a service-chain with 3 service-instances and 2 SVMs in each of the SI between vn1 and vn2.
2]. Observed that ping from vm1 to vm2 fails.
Naveen is aware of the issue.
information type: | Proprietary → Public |
tags: | added: releasenote |
To post a comment you must log in.
Reviewed: https:/ /review. opencontrail. org/3817 github. org/Juniper/ contrail- controller/ commit/ a21d82accf1b38e c5597feb5c304c5 d40f104359
Committed: http://
Submitter: Zuul
Branch: R1.10
commit a21d82accf1b38e c5597feb5c304c5 d40f104359
Author: Naveen N <email address hidden>
Date: Thu Oct 16 04:43:55 2014 -0700
* All the component nh member of a composite NH are policy disabled
nexthop, since composite NH itself is policy enabled and there is
no need for component member to policy disabled. When a transition
happens from non-ecmp to ecmp, local interface nexthop could be
added with policy enabled flag, and finding of the nexthop inside
composite NH would fail since we always search for policy disabled
nh inside composite NH. Since nexthop lookup fails flow ecmp index
would not be set, resulting in drop
Unit test added to verify the case where transition happens from
non-ecmp to ecmp with local nexthop pre-existing
Closes-bug:#1381978
Change-Id: Ib097a77450afcc 7ebba622a0732af 3c1105c7db9