Let's say the following is in the HPEL or SystemOut log when attempting to start your WebSphere network deployment manager (dmgr), node, or application server.

CWRLS0030W: Waiting for HAManager to activate recovery processing for local WebSphere server.

DCSV8030I

Check the SystemOut.log or HPEL log for event DCSV8030I.

DCSV8030I: DCS Stack DefaultCoreGroup at Member cell\node\appserver001: 
Failed to join or establish a view with member [cell\mgr\dmgr]. 
The reason is Not all candidates are connected ConnectedSetMissing =  [ ] ConnectedSetAdditional [ cell\node\appserver002].

If the SystemOut.log contains event DCSV8030I, according to this IBM article, the cause of this problem is that two or more application servers “disagree with each others member connection list”. In this scenario, the application servers that disagree are appserver001 and appserver002.

The IBM article suggests that all members (aka application servers) should be stopped and then restarted. Typically, stopping both application servers and then starting each application server one at a time will resolve this issue. After stopping both application servers, use the ps command (on Linux) to ensure there are no processes associated with either application server.

ps -ef | grep appserver001
ps -ef | grep appserver001

If there are any active processes with either application server, use the kill command (on Linux) to kill the processes.

kill -9 the_pid

In some scenarios, you may need to start application server "a" first, and then application server "b" second. Or, you may need to start application server "b" first, and then application server "a" second.

Allow DCS_UNICAST_ADDRESS in firewall

You may also need to allow the DCS_UNICAST_ADDRESS port in your firewall, such as iptables or firewalld (on Linux). Refer to IBM WebSphere - Firewall to determine the DCS_UNICAST_ADDRESS port.