[AMBARI-15330] Bubble up errors during RU/EU

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.0
• Fix Version/s: 2.4.0
• Component/s: ambari-server
• Assignee: Alejandro Fernandez
• Reporter: Alejandro Fernandez
• Created: 08/Mar/16 02:04
• Updated: 11/Mar/16 20:31
• Resolved: 11/Mar/16 20:31

Description

During RU/EU, need a way to bubble up an error of the current item that failed. This is useful to quickly get a human-readable error that others UIs can quickly retrieve.
It can print a human-readable error, plus stdout and stderr.

This would become part of the upgrade endpoint. e.g,
api/v1/clusters/$name/upgrade_summary/$request_id

{
attempt_cnt: 1,
cluster_name: "c1",
request_id: 1,
fail_reason: "Failed calling RESTART ZOOKEEPER/ZOOKEEPER_SERVER on host c6401.ambari.apache.org",

// Notice that the rest are inherited from the failed task if it exists.
command: "CUSTOM_COMMAND",
command_detail: "RESTART ZOOKEEPER/ZOOKEEPER_SERVER",
custom_command_name: "RESTART",
end_time: -1,
error_log: "/var/lib/ambari-agent/data/errors-1234.txt",
exit_code: 1,
host_name: "c6401.ambari.apache.org",
id: 1234,
output_log: "/var/lib/ambari-agent/data/output-1234.txt",
role: "ZOOKEEPER_SERVER",
stage_id: 1,
start_time: 123456789,
status: "HOLDING_FAILED",
stdout: "",
stderr: ""
}

Issue Links

• links to: Review Board patch