This documentation will show you have to update PureInsight Cluster’s bound hostname. This will bind the services associated with the PureInsight Cluster.

You need to bind to a hostname rather than IP.

PreReqs

You’ll need to update PureInsight to at least 2.1.0
Identify the network layout and determine IP and hostnames to use
Turn off Alerts, IDS, Pcap Indexer, and realtime stats

Steps

You’ll want to be on the QP that you’ve updated the hostname of.
First update the hostname on the QP to whatever you want.
1. You will have to go into root to set the hostname this way
Make sure you’re actually able to connect to this QP from all the other QP’s within the cluster after the hostname has been updated
1. This might involve altering each /etc/hosts file if no DHCP or DNS server is set up
Change into the admin user su admin
Run the bash script bash /opt/quantea/api/pi_cluster_management/pi_cluster_setup.sh update_hostname
1. Follow the prompts
  1. If running the script for the first time - you’ll need to udpate the SSL Certificates. Type in y
  2. You’ll need to update the SANS file
    1. Add the new hostname(s) using DNS.2=hostname. You can bind as many hostnames or IP’s as you want. DNS.X or IP.X
  3. Make sure to verify your host’s keys
2. This script will automatically detect what roles this QP has within the cluster.
3. Mongo Role
  1. This script will prompt you to ask which hostname you’re updating.

Select the appropriate member that you’re updating

d. Kafka Role: This script will go ahead and reconfigure Kafka automatically

The script will then continue, and set up the connectivity on each QP for the new node. It may prompt you to verify the new node’s host key - since the previous host key is under the previous hostname.
1. You’ll have to verify the key is correct
After it’s done you can verify Kafka and Mongo again
1. python pi_cluster_management/kafka_cluster_management.py --verify_kafka
2. python pi_cluster_management/mongo_cluster_management.py --verify

Errors

No host described in new configuration

A common error that occurs during the update process is

An operation failure that says something along the lines of No host described in new configuration with {version: 2, term: 2} for replica set piRS13375646 maps to this node, full error: {'ok': 0.0, 'errmsg': 'No host described in new configuration with {version: 2, term: 2} for replica set piRS13375646 maps to this node

It is due to the configuration file being updated with the new host, and the host being rejected. This is more common in replica sets that don’t maintain availability or with a replica set with only one host.

All you have to do is to restart the node sudo systemctl restart mongod and then run the script again.

And we’ll see that the successive run of the update_hosts script works.

Temporary failure in name resolution

This error will occur after setting up known hosts on each of the units within the cluster after the hostname has been set. This usually happens if you did not verify that all units could connect to the new hostname

We can see in this picture that we successfully updated the hostname of qp4k-pidev01 to qp4k-pidev191.

However when setting up the connectivity for all the other hosts, we can see that we can’t connect to qp4k-pidev01 with the updated hostname qp4k-pidev191