Installation stuck at "Setting up firewall…​" step

Problem

Installation was stuck at “Setting up firewall…” step, and eventually failed after about ~2 hours.

Error message

[PROGRESS]: 13:33:38 Setting up firewall...
[PROGRESS]: 13:33:38 Setup the firewall rules on m1 (10.174.36.11) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m1
[PROGRESS]: 13:33:38 Setup the firewall rules on m2 (10.174.47.133) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m2
[PROGRESS]: 13:33:38 Setup the firewall rules on m3 (10.174.47.136) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m3
[PROGRESS]: 13:33:38 Setup the firewall rules on m4 (10.174.47.132) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m4
[PROGRESS]: 13:33:38 Setup the firewall rules on m5 (10.174.47.134) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m5
[PROGRESS]: 13:33:38 Setup the firewall rules on m6 (10.174.47.135) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m6
[PROGRESS]: 13:33:38 Setup the firewall rules on m7 (10.174.47.129) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m7
[PROGRESS]: 13:33:38 Setup the firewall rules on m8 (10.174.47.130) in background, you may check the log at:
/u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/setup_firewall.log.m8
[PROGRESS]: 13:33:38 Wait until setup firewall on each node to finish, this may take a while ...
[PROGRESS]: 15:32:15 Masking firewall
[PROGRESS]: 15:32:15 Checking ports access in the cluster...
------------------------------------------------------------
[NOTE    ]: Node m1 (10.174.36.11) port check passed
[ERROR   ]: Node m2 (10.174.47.133) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m2
[ERROR   ]: Node m3 (10.174.47.136) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m3
[ERROR   ]: Node m4 (10.174.47.132) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m4
[ERROR   ]: Node m5 (10.174.47.134) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m5
[ERROR   ]: Node m6 (10.174.47.135) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m6
[ERROR   ]: Node m7 (10.174.47.129) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m7
[ERROR   ]: Node m8 (10.174.47.130) cannot access the required ports of one or more cluster nodes, please check the log: /u06/tgh-shared/software/tigergraph-3.5.2-offline/logs/ports_check.log.m8
[ERROR   ]: The check_ports on one or more nodes failed

Diagnosis

As the installation is being run from the m1 node, it needs to be able to access the other remaining nodes within the cluster. The errors above are a good indication that the connection between m1 and the other nodes is not being established.

Ensure that ssh is enabled on all nodes. From the m1 node, you can run the following command to test ssh connectivity to the remaining m* nodes:

ssh -q <hostname_for_m*_node>

Workaround

One workaround is to run the ../<tigergraph_package_dir>/utils/setup_firewall_on_nodes.sh script manually on each node.

From there, you can retry the TigerGraph installation script.

Solution

Solution is to enable ssh on all nodes within the cluster.