Remove a Node From Carbonio Infrastructure

Remove a Node From Carbonio Infrastructure#

There are a number of scenarios in which an existing Carbonio Node must be removed from a Carbonio infrastructure within an organisation, for example:

Lifecycle Completion: The VM hosting the Node has fulfilled its purpose, such as completing a specific task, project, or testing phase.
Resource Optimisation: Free up computing resources like CPU, memory, and storage that can be used for other purposes or tasks within an organisation.
Costs: In cloud environments, disposing of unused or underutilised VMs reduces costs.
Obsolescence: The Operating system or applications running on the VM are outdated or no longer supported.

Preliminaries#

Before starting the procedure described in the remainder of this section, you need to:

Have CLI access to the new Node and to the Nodes featuring the Directory Server and the Mailstore & Provisioning Roles
Identify which Node (or Nodes) should be removed or replaced and which Roles they are equipped with
Install these Roles on other Nodes
Take note of the FQDN of the Node to be removed, because it will be needed. It will be denoted as serverHostname in the procedure
in case you remove a Proxy Role or A Mailstore & Provisioning Role, check that the Preview Role is correctly configured. See Section Configuration Fix for Proxy Removal and Configuration Fix for Mailstore & Provisioning Removal, respectively, for directions.

Node Removal#

You can now start the removal procedure. If you plan to remove more than one Node, execute the following steps for each Node.

Step 1, leave Carbonio Mesh

On the Node that will be dismissed, generate a Carbonio Mesh token:

# export CONSUL_HTTP_TOKEN=$(gpg -qdo - \
/etc/zextras/service-discover/cluster-credentials.tar.gpg | \
tar xOf - consul-acl-secret.json | jq .SecretID -r)

Remove the Node from the cluster:

# consul leave

Step 2, remove the Node from the Carbonio infrastructure

Log in to the Node on which the Mailstore & Provisioning Role is installed and remove the server from the infrastructure. As the zextras user, execute the command

zextras$ carbonio prov ds serverHostname

Step 3, verify removal

Log in to the Node equipped with the Directory Server Role and dump the LDAP main database: as the zextras user, execute the command

zextras$ /opt/zextras/libexec/zmslapcat /tmp/

The output will be file /tmp/ldap.bak; check that the file does no longer contain any reference to the old FQDN, i.e.,

zextras$ grep serveHostname /tmp/ldap.bak

Hint

The command can be run as either the root or zextras user.

The output of the command should be empty, meaning that the old Node is not part of the Carbonio infrastructure anymore. You can now power off the Node and decommission it.

Configuration Fix for Proxy Removal#

In case you are removing from the infrastructure a Node which installs a Proxy, you need to adjust a configuration file on the Preview Node.

This is required by the memcached configuration that supports the Preview Role and consists of replacing a configuration value with the IP address of the new Proxy Role.

The procedure is slight different if you if you replace an existent Proxy with a new one or simply decommission one Proxy but keep an existing one (or more that one).

Replace a Proxy Node

Supposing that the IP of the decommissioned Proxy Node is 172.16.0.12, and the new one is 172.16.0.73, you need to change in file /etc/carbonio/preview/config.ini the line

memcached_server_full_path_urls = 172.16.0.12:11211

into

memcached_server_full_path_urls = 172.16.0.73:11211

Do not replace a Proxy Node

In case you decommission a Proxy Node without replacing it, you still need to edit file /etc/carbonio/preview/config.ini, but you need to delete the IP of the decommissioned Proxy Node, for example (assuming you are decommissioning 172.16.0.12)

memcached_server_full_path_urls = 172.16.0.12:11211,172.16.0.22:11211

becomes

memcached_server_full_path_urls = 172.16.0.22:11211

Please refer to Section Configure Memcached in Preview Role’s installation for details.

Configuration Fix for Mailstore & Provisioning Removal#

In case you are removing from the infrastructure a Node which installs a Mailstore & Provisioning, you need to adjust a configuration file on the Preview Node.

This is required by the memcached configuration that supports the Mailstore & Provisioning Role and consists of replacing a configuration value with the IP address of the new Mailstore & Provisioning Role.

The procedure is slight different if you if you replace an existent Mailstore & Provisioning with a new one or simply decommission one Mailstore & Provisioning but keep an existing one (or more that one).

Replace a Mailstore & Provisioning Node

Supposing that the IP of the decommissioned Mailstore & Provisioning Node is 172.16.0.13, and the new one is 172.16.0.58, you need to change in file /etc/carbonio/preview/config.ini the line

nginx_lookup_server_full_path_urls = https://172.16.0.13:7072