[AMBARI-5722] All Services Fail To Deploy Due To Agent Parsing Exception

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 1.6.0
• Fix Version/s: 1.6.0

Environment

OSX (Ambari Server)
CentOS 6.4 (Ambari Agents)
The server must be deployed on a machine that is not part of the cluster to manifest this error.

• Assignee: Dmitry Lysnichenko
• Reporter: Jonathan Hurley
• Created: 09/May/14 21:10
• Updated: 12/May/14 15:03
• Resolved: 12/May/14 15:02

Description

When deploying a brand new cluster, all services fail to install due to a parsing exception thrown from the Ambari Agents.
File "/usr/lib/python2.6/site-packages/ambari_agent/CustomServiceOrchestrator.py", line 113, in runCommand
json_path = self.dump_command_to_json(command)
File "/usr/lib/python2.6/site-packages/ambari_agent/CustomServiceOrchestrator.py", line 209, in dump_command_to_json
command'clusterHostInfo' = manifestGenerator.decompressClusterHostInfo(command'clusterHostInfo')
File "/usr/lib/python2.6/site-packages/ambari_agent/manifestGenerator.py", line 116, in decompressClusterHostInfo
indexes = convertRangeToList(v)
File "/usr/lib/python2.6/site-packages/ambari_agent/manifestGenerator.py", line 57, in convertRangeToList
raise AgentException.AgentException("Broken data in given range, expected - ""m-n"" or ""m"", got : " + str(r))
AgentException: 'Broken data in given range, expected - m-n or m, got : -1’
The command being sent is

{hs_host=[2], namenode_host=[1], snamenode_host=[2], zookeeper_hosts=[0-2], ganglia_server_host=[1], nm_hosts=[0], ganglia_monitor_hosts=[0-2], all_hosts=[c6403.ambari.apache.org, c6401.ambari.apache.org, c6402.ambari.apache.org], rm_host=[2], app_timeline_server_hosts=[2], slave_hosts=[0], ambari_server_host=[-1], nagios_server_host=[1], all_ping_ports=[8670:0-2]}

Notice the ambari-server-host which was added in that commit; it’s value is “-1” which would not parse correctly in manifestGenerator.py
I suspect Git e667dc7c9870864ff537374c819b7c1d1dd88e98 caused this problem.
Steps to reproduce:
1) Provision 3 c64 hosts
2) Wipe your server database and re-create it with the embedded PSQL script
3) Attempt to provision a cluster with various services.
All services will fail to deploy b/c of the above exception. This was working without issues before the above suspect commit.

Issue Links

• links to: Review board (with fix)