[AMBARI-21468] Knox restart fails during EU due to wrong location to the Knox data directory

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: branch-feature-AMBARI-21348
• Fix Version/s: branch-feature-AMBARI-21348
• Component/s: ambari-server
• Epic Link: Cross Stack Name Upgrades
• Assignee: Di Li
• Reporter: Di Li
• Created: 13/Jul/17 16:25
• Updated: 13/Jul/17 16:39
• Resolved: 13/Jul/17 16:39

Description

Knox fails the restart during EU with error
"""
resource_management.core.exceptions.ExecutionFailed: Execution of 'cp -R -p -f /usr/hdp/4.2.0.0/knox/data/. /usr/hdp/current/knox-server/data' returned 1. cp: cannot stat '/usr/hdp/4.2.0.0/knox/data/.': No such file or directory
"""

This is because the common service Knox Py script assumes data dir is under stack_root which is /usr/hdp by the time it's Knox to restart during EU. But it also attaches the "current_version" obtained from command json file that points to 4.2.0.0 (or 4.2.5.0-0000) for the migration cluster. So the result root dir /usr/hdp/4.2.xxxx does not exist.

"""
ambari-server\src\main\resources\common-services\KNOX\0.5.0.2.2\package\scripts\upgrade.py

1. <stack-root>/2.3.0.0-1234/knox/data/.
   source_data_dir = os.path.join(params.stack_root, params.upgrade_from_version, "knox", "data", ".")

ambari-server\src\main\resources\common-services\KNOX\0.5.0.2.2\package\scripts\params_linux.py
upgrade_from_version = default("/hostLevelParams/current_version", None)
"""

"""
"hostLevelParams": {
"agent_stack_retry_on_unavailability": "false",
"stack_name": "HDP",
"group_list": "[\"spark\",\"hadoop\",\"users\",\"knox\"]",
"host_sys_prepped": "false",
"ambari_db_rca_username": "mapred",
"current_version": "4.2.0.0",
"""

Issue Links

• duplicates: AMBARI-21466 Knox upgrade fails due to wrong stack root