HDP 2.4.0 Installation with Ambari - No progress at Step 3 / Confirm Hosts

[#1]

Hello,

I am trying to install HDP 2.4 using Ambari 2.2.2.0 on Ubuntu14.

FQDN and passwordless SSH works fine, hostname (-f) is displayed as expected and I can login as root using ssh without password. When I try to deploy the cluster - there is no progress at the Step 3: Confirm Hosts.

Because no ambari agent is running on both hosts in the cluster, I installed, configured and started them manually as described in the documentation. Registration with ambari server was succesful and heartbeat is received by both hosts. Here is one of the agent logs:

INFO 2016-06-29 15:07:51,932 logger.py:71 - call returned (0, '')
INFO 2016-06-29 15:07:51,932 logger.py:71 - call['test -w /run/user'] {'sudo': True, 'timeout': 5}
INFO 2016-06-29 15:07:51,940 logger.py:71 - call returned (0, '')
INFO 2016-06-29 15:07:51,940 logger.py:71 - call['test -w /boot/efi'] {'sudo': True, 'timeout': 5}
INFO 2016-06-29 15:07:51,947 logger.py:71 - call returned (0, '')
INFO 2016-06-29 15:07:51,952 Controller.py:265 - Heartbeat response received (id = 919)
INFO 2016-06-29 15:08:01,853 Heartbeat.py:78 - Building Heartbeat: {responseId = 919, timestamp = 1467205681853, commandsInProgress = False, componentsMapped = False}
INFO 2016-06-29 15:08:01,858 Controller.py:265 - Heartbeat response received (id = 920)

I then retried to deploy the cluster but there is still no progress. When I look at the ambari server log file there is a constant loop of this message

29 Jun 2016 12:52:41,225  INFO [pool-9-thread-1] BSHostStatusCollector:55 - Request directory /var/run/ambari-server/bootstrap/1
29 Jun 2016 12:52:41,225  INFO [pool-9-thread-1] BSHostStatusCollector:62 - HostList for polling on [webprojekttemp-2.f07-imp.fh-koeln.de, webprojekttemp-3.f07-imp.fh-koeln.de]
29 Jun 2016 12:52:41,802  INFO [pool-10-thread-1] BSHostStatusCollector:55 - Request directory /var/run/ambari-server/bootstrap/2
29 Jun 2016 12:52:41,802  INFO [pool-10-thread-1] BSHostStatusCollector:62 - HostList for polling on [webprojekttemp-2.f07-imp.fh-koeln.de, webprojekttemp-3.f07-imp.fh-koeln.de]
29 Jun 2016 12:52:43,012  INFO [pool-11-thread-1] BSHostStatusCollector:55 - Request directory /var/run/ambari-server/bootstrap/3
29 Jun 2016 12:52:43,012  INFO [pool-11-thread-1] BSHostStatusCollector:62 - HostList for polling on [webprojekttemp-2.f07-imp.fh-koeln.de, webprojekttemp-3.f07-imp.fh-koeln.de]

The directory

/var/run/ambari-server/bootstrap/

does not exist. How to deal with that problem?

[#2]

@Johannes Gi, make sure that there is not a firewall running on any of the hosts. You can ssh in but the ports are different for agent server communication. Check the agent logs as well.

[#3]

Hey @Ian Roberts

On the cluster hosts, iptables is configured that out and incoming traffic to / from the ambari server is accepted on ALL ports. Here is the agent-log of one host, the other log looks the same:

...
INFO 2016-06-29 15:07:51,932 logger.py:71 - call returned (0, '')
INFO 2016-06-29 15:07:51,932 logger.py:71 - call['test -w /run/user'] {'sudo': True, 'timeout': 5}
INFO 2016-06-29 15:07:51,940 logger.py:71 - call returned (0, '')
INFO 2016-06-29 15:07:51,940 logger.py:71 - call['test -w /boot/efi'] {'sudo': True, 'timeout': 5}
INFO 2016-06-29 15:07:51,947 logger.py:71 - call returned (0, '')
INFO 2016-06-29 15:07:51,952 Controller.py:265 - Heartbeat response received (id = 919)
INFO 2016-06-29 15:08:01,853 Heartbeat.py:78 - Building Heartbeat: {responseId = 919, timestamp = 1467205681853, commandsInProgress = False, componentsMapped = False}
INFO 2016-06-29 15:08:01,858 Controller.py:265 - Heartbeat response received (id = 920)

[#4]

Since you installed ambari-agent manually, Have you selected "Perform manual registration" option on previous page

[#5]

I am also facing same problem.let me know once you fixed this issue please

[#6]

Out of curiosity, are the nodes (that you are trying to add) able to communicate back with the Ambari server? Have you verified the DNS resolution (with the DNS name found in /etc/ambari-agent/conf/ambari-agent.ini file)? Are the required ports allowed on the Ambari server side (8440)?

[#7]

I am using SSH passwordless method. so i am not doing manually Ambari-agent so that agent.ini file is not available.and 8440 port also enabled

[#8]

Manually starting the agents and registration without SSH did the job. But starting the services failed after installation: 62 alerts. I want to revert the deployment... Any suggestion how to do that the simplest and cleanest way to deploy again? I selected all services in the previous try...