You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@impala.apache.org by "Tianyi Wang (JIRA)" <ji...@apache.org> on 2018/04/13 20:38:00 UTC

[jira] [Created] (IMPALA-6851) custom cluster test TestGrantRevoke::test_role_update doesn't clean up restarted impala daemons correctly

Tianyi Wang created IMPALA-6851:
-----------------------------------

             Summary: custom cluster test TestGrantRevoke::test_role_update doesn't clean up restarted impala daemons correctly
                 Key: IMPALA-6851
                 URL: https://issues.apache.org/jira/browse/IMPALA-6851
             Project: IMPALA
          Issue Type: Bug
          Components: Infrastructure
    Affects Versions: Impala 2.12.0
            Reporter: Tianyi Wang
            Assignee: Tianyi Wang
             Fix For: Impala 3.0


In TestGrantRevoke::test_role_update, one of the impala daemons is restarted. When the test finishes, it doesn't kill or wait for the restarted process. There are several problems of not doing so:
- When another custom cluster test attempts to kill the impala daemon, the impala daemon will become a zombie process. The PID won't go away until the py tests process exits. start-impala-cluster.py doesn't handle this case correctly and will time out. Fortunately, test_role_update is always the last test in custom cluster tests, so it doesn't break builds for now.
- It leaks PID and process information in the kernel as long as the py test process is running.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)