Dave Beckett created AMBARI-11952:
-------------------------------------
Summary: get_mount_point_for_dir does not always calculate the
right mount point
Key: AMBARI-11952
URL: https://issues.apache.org/jira/browse/AMBARI-11952
Project: Ambari
Issue Type: Bug
Components: ambari-agent
Affects Versions: 2.0.1
Reporter: Dave Beckett
Priority: Minor
get_mount_point_for_dir() mount checking code calculates that the best disk
partition for a data directory /diskN/hadoop/dfs under a disk mounted at /diskN
as / rather than the expected answer /diskN.
This is because the number of separators (slashes) in '/' is 1 and the number
of them in '/diskN' is also 1, and root tends to be first in mount(1) output,
so it wins.
This code here should count the number of non-None path segments, not the
number of separators. In python terms, the name with the maximum len([x for x
in best_mount_found.split(os.path.sep) where x is not None])
This was found while researching AMBARI-11951
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
