Page tree

Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Add a compute node

You already deployed successfully Joid and want to scale up ? this is how to add a compute node:

  • Add a new compute node unit:
Code Block
languagebash
themeFadeToGrey
  juju add-unit nova-compute 

 

  • Check the new compute is ready
Code Block
languagebash
themeFadeToGrey
  juju status --format=tabular | grep nova-compute

 

  • Get the ID of your last added compute and add Ceph to this node:
Code Block
languagebash
themeFadeToGrey
  ID=$(juju status --format=tabular | tail -n2 |head -n1 | cut -d\  -f1)
  juju add-unit ceph-osd --to $ID

 

  • Check ceph is ready
Code Block
languagebash
themeFadeToGrey
  juju status --format=tabular | grep ceph-osd

Table of Contents

Update CEPH number of replicat

Code Block
languagebash
themeFadeToGrey
  juju set ceph-radosgw ceph-osd-replication-count=2

 

Configure

NAT for

access to Public API on HA mode

There is 2 way of configuring public access on API:

  • Production ready: Using a NAT between a public IP and the admin network
  • Test/CI usage: using a reverse proxy on the jumphost

Using a NAT

Register API FQDN

First set your DNS:

Code Block
languagetext
titleDNS Entries
api               28800  A <PUBLIC_IP_TO_NAT>
heat              28800  CNAME  api
glance            28800  CNAME  api
cinder            28800  CNAME  api
ceilometer        28800  CNAME  api
nova              28800  CNAME  api
neutron           28800  CNAME  api
swift             28800  CNAME  api

Set NAT rules

Set those NAT rules on the default router:

namepublic portinternal IPinternal port
api

5000

<ADMIN_NETWORK_VIP_PREFIX>3

5000

heat8004<ADMIN_NETWORK_VIP_PREFIX>88004
glance9292<ADMIN_NETWORK_VIP_PREFIX>29292
heat8000<ADMIN_NETWORK_VIP_PREFIX>88000
cinder8776<ADMIN_NETWORK_VIP_PREFIX>98776
nova8774<ADMIN_NETWORK_VIP_PREFIX>68774
neutron9696<ADMIN_NETWORK_VIP_PREFIX>79696
swift80<ADMIN_NETWORK_VIP_PREFIX>180
ceilometer8777<ADMIN_NETWORK_VIP_PREFIX>48777

<ADMIN_NETWORK_PREFIX> is the prefix of the virtual ips of your clusters. As an example, <ADMIN_NETWORK_PREFIX>3 on the Orange pod2 will be 192.168.2.23

Test/CI usage: using a reverse proxy

Add a public IP on the jumphost

Add an IP from your JOID public network on the jumphost (the next after the router ip for example)

Set the reverse proxy

Edit joid/ci/joid_pub_access_nginx.sh

Code Block
languagebash
themeFadeToGrey
firstline33
linenumberstrue
# Public Network (to authorize only the admin network to connect api
PUBNET=10.0.0.0/24
# Public IP
PUBIP=10.0.1.3
# Public interface
PUBIF=eth1.120
# Admin interface
ADMIF=brAdm
# Admin network prefix
ADMNETPREF=192.168.1.2
# Domain name for APIs
DOMAIN=pod1.opnfv.fr

Run the script:

Code Block
languagebash
themeFadeToGrey
bash joid/ci/joid_pub_access_nginx.sh

To delete the reverse proxy and clean iptables rules, run:

Code Block
languagebash
themeFadeToGrey
bash joid/ci/joid_pub_access_nginx.sh -D