[jira] [Created] (OOZIE-3354) [core] [action] SSH action gets hung

["Andras Piros (JIRA)" <ji...@apache.org>]

Andras Piros created OOZIE-3354:
-----------------------------------

             Summary: [core] [action] SSH action gets hung
                 Key: OOZIE-3354
                 URL: https://issues.apache.org/jira/browse/OOZIE-3354
             Project: Oozie
          Issue Type: Bug
          Components: action, core
    Affects Versions: 5.0.0
            Reporter: Andras Piros
            Assignee: Andras Piros
             Fix For: 5.1.0


In OOZIE-3183 {{SshActionExecutor#drainBuffers()}} has changed. Previously, it called {{Process#exitCode()}} that would return immediately either with the exit code, or would throw an {{IllegalThreadStateException}} if the process would still be running.

In the current implementation introduced by OOZIE-3183, {{Process#waitFor()}} is used that would block until the process finishes. Given the fact that sometime {{SshActionExecutor#check()}} calls {{ssh ... cat stdout}}, and this SSH process can be trapped even after {{cat stdout}} has been finished on the target host, it can happen that {{SshActionExecutor#drainBuffers()}} waits indefinitely without a chance to gather any {{stdout}} or {{stderr}} logs. Hence this particular one is a compatibility breaking change with existing SSH action behavior.

Let's re-introduce the former behavior in {{SshActionExecutor#drainBuffers()}} that keeps polling {{Process#exitValue()}} and reading the progress on {{stdout}} and {{stderr}} till the process finishes, for backwards compatibility.

[This article|https://www.javaworld.com/article/2071275/core-java/when-runtime-exec---won-t.html] covers the behavioral details of {{Process#waitFor()}}.



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