vcenter-HA provision failed due to missing nfs package in contrail-vm(compute)

[shajuvk]

Hi Team,

HA configuration in Vcenter as compute setup failed with the below error during setup_all. If we set contrail-VM as the first compute in testbed.py we will get into this issue.

As apart of install vrouter, we are setting Manage_nova= NO, due to this installing nfs packages which is need on vrouter not installed.
In HA setup we need a common location for glance images.

root@ContrailVM-b5s41:/opt/contrail/contrail_install_repo# ls -l nfs*
-rwxr-xr-x 1 33694 757 180760 Dec 16 21:31 nfs-common_1%3a1.2.8-6ubuntu1.2_amd64.deb
-rwxr-xr-x 1 33694 757 85826 Dec 16 21:31 nfs-kernel-server_1%3a1.2.8-6ubuntu1.2_amd64.deb

Error log:
=======
2016-02-05 10:36:02:324227: [root@10.84.24.231] sudo: echo "/var/tmp/glance-images *(rw,sync,no_subtree_check)" >> /etc/exports
2016-02-05 10:36:02:423240: [root@10.84.24.231] sudo: sudo /etc/init.d/nfs-kernel-server restart
2016-02-05 10:36:02:489965: [root@10.84.24.231] out: sudo: /etc/init.d/nfs-kernel-server: command not found
2016-02-05 10:36:02:556631: [root@10.84.24.231] out:
2016-02-05 10:36:02:913790: Disconnecting from 10.84.24.231... done.
2016-02-05 10:36:03:187451:
2016-02-05 10:33:08:587405: Warning: sudo() received nonzero return code 1 while executing 'ip addr | grep 192.168.10.201'!
2016-02-05 10:33:08:587405:
2016-02-05 10:33:08:587405:
2016-02-05 10:36:02:557324: Fatal error: sudo() received nonzero return code 1 while executing!
2016-02-05 10:36:02:557324:
2016-02-05 10:36:02:557324: Requested: sudo /etc/init.d/nfs-kernel-server restart
2016-02-05 10:36:02:557324: Executed: sudo -S -p 'sudo password:' /bin/bash -l -c "sudo /etc/init.d/nfs-kernel-server restart"
2016-02-05 10:36:02:557324:
2016-02-05 10:36:02:557442: Aborting.
2016-02-05 10:36:02:557442:

Thanks,
Shaju

[shajuvk (shajuvk)]

As a work around, if we specify KVM compute as the first router in testbed file this issue won't be there

[shajuvk (shajuvk)]

As a work around, if we specify KVM compute as the first compute in testbed file we won’t hit this issue

[shajuvk (shajuvk)]

please check the multi node example.. Need nfs configuration in testbed file to over come this issue in vcenter as compute or orchstrator setup.

specify nfs server and nfs server path for mounting location for glance images

example :
======
 nfs_server : NFS server to be used to store the glance images.
 nfs_glance_path : NFS server image path, which will be mounted on the Openstack Nodes and
# the glance images will be placed/accesed in/from this location.
#env.ha = {
# 'internal_vip' : '1.1.1.100', #Internal Virtual IP of the openstack HA Nodes.
# 'external_vip' : '2.2.2.200', #External Virtual IP of the openstack HA Nodes.
# 'contrail_internal_vip' : '1.1.1.10', #Internal Virtual IP of the contrail HA Nodes.
# 'contrail_external_vip' : '2.2.2.20', #External Virtual IP of the contrail HA Nodes.
   'nfs_server' : '3.3.3.3', #IP address of the NFS Server which will be mounted to /var/lib/glance/images of openstack Node, Defaults to env.roledefs['compute'][0]
 'nfs_glance_path' : '/var/tmp/images/', #NFS Server path to save images, Defaults to /var/tmp/glance-images/
#}

[Sachchidanand Vaidya (vaidyasd)]

As per the discussion, in HA setup, we will change the location of glance database to nfs location rather than using ContrailVM as the glance image store.