Bug #1725461 “Contrail 3.1 : Need to upgrade the keepalived pack...” : Bugs : Juniper Openstack

vivekananda shenoy (vshenoy83) on 2017-10-20

Changed in juniperopenstack:
assignee:	nobody → amudhar (amudha)
milestone:	none → r3.1.4.0

vivekananda shenoy (vshenoy83) on 2017-10-21

information type:	Proprietary → Public
description:	updated

Revision history for this message

vivekananda shenoy (vshenoy83) wrote on 2017-10-24:

#1

Download full text (10.1 KiB)

Hi Amudha,

This is the complete description of keepalived issue as described by NTTC. Please make sure that whatever keepalived version we are planning to use in the fixed build has the fix for this issue. Just want to be double sure.

===================

Hi Toby-san,

As for issue 1, I tested this in my lab, and confirmed VRRP flood was caused by the wrong definition in testbed.py like below.

----------------------------------------------------------------------------

env.roledefs = {

'all': [host1, host2, host3, host4, host5, host6, host7, host8, host9],

'cfgm': [host1, host2, host3],

'control': [host1, host2, host3],

'webui': [host1, host2, host3],

'collector': [host4, host5, host6],

'database': [host7, host8, host9],

'openstack': [openstack],

'compute': [tsn1, tsn2, compute1, compute2],

'tsn': [tsn1, tsn2],

'toragent': [tsn1, tsn2],

'build': [host_build],

:

}

control_data = {

#host1: {'ip': '192.168.0.131/24', 'gw': '192.168.0.1', 'device': 'eth1'}, <<< commented out

#host2: {'ip': '192.168.0.132/24', 'gw': '192.168.0.1', 'device': 'eth1'}, <<< commented out

#host3: {'ip': '192.168.0.133/24', 'gw': '192.168.0.1', 'device': 'eth1'}, <<< commented out

host4: {'ip': '192.168.0.134/24', 'gw': '192.168.0.1', 'device': 'eth1'},

host5: {'ip': '192.168.0.135/24', 'gw': '192.168.0.1', 'device': 'eth1'},

host6: {'ip': '192.168.0.136/24', 'gw': '192.168.0.1', 'device': 'eth1'},

host7: {'ip': '192.168.0.137/24', 'gw': '192.168.0.1', 'device': 'eth1'},

host8: {'ip': '192.168.0.138/24', 'gw': '192.168.0.1', 'device': 'eth1'},

host9: {'ip': '192.168.0.139/24', 'gw': '192.168.0.1', 'device': 'eth1'},

:

}

env.ha = {

'contrail_internal_vip': '192.168.0.110',

'contrail_external_vip': '10.0.0.110',

'contrail_internal_virtual_router_id': 110,

'contrail_external_virtual_router_id': 110,

}

----------------------------------------------------------------------------

With this testbed.py, setup_contrail_keepalived task passed wrong arguments to setup-vnc-keepalived.

Usually, self_ip and internal_vip address must be on the same network, but different with this case.

----------------------------------------------------------------------------

root@sv-31:/opt/contrail/utils# grep -E "setup_contrail_keepalived|setup-vnc-keepalived" setup_without_openstack_and_compute_2017_10_21_13_54_06_540732.log

2017-10-21 13:54:14:352328: [root@10.0.0.131] Executing task 'setup_contrail_keepalived'

2017-10-21 13:54:15:068439: [root@10.0.0.131] sudo: setup-vnc-keepalived --self_ip 10.0.0.131 --internal_vip 192.168.0.110 --mgmt_self_ip 10.0.0.131 --self_index 1 --num_nodes 3 --role cfgm --internal_virtual_router_id 110 --external_vip 10.0.0.110 --external_virtual_router_id 110

2017-10-21 13:54:16:991731: [root@10.0.0.132] Executing task 'setup_contrail_keepalived'

2017-10-21 13:54:19:794308: [root@10.0.0.132] sudo: setup-vnc-keepalived --self_ip 10.0.0.132 --internal_vip 192.168.0.110 --mgmt_self_ip 10.0.0.132 --self_index 2 --num_nodes 3 --role cfgm --internal_virtual_router_id 110 --external_vip 10.0.0...