Sorry that I have not provided relevant info about the problem,now i will try
to describe this problem clearly because it is a little confusing.
When there are 3 or more this unique workflow running simultaneously(actually
in prep status instead of running status,i guess custom action implemented by
custom runs its code in prep status ), any other workflow(has nothing to do
with the unique wf) can not run,always hang on PREP status.
And the 3 or more unique workflow can not be killed ,the oozie.log shows I
have killed it many times but the kill command does not work.
2013-11-11 07:44:02,449 INFO DagEngine:539 - USER[-] GROUP[-] TOKEN[-] APP[-]
JOB[-] ACTION[-] User hadoop killed the WF job
0000001-131111070843260-oozie-hado-W
2013-11-11 07:44:12,458 INFO DagEngine:539 - USER[-] GROUP[-] TOKEN[-] APP[-]
JOB[-] ACTION[-] User hadoop killed the WF job
0000001-131111070843260-oozie-hado-W
2013-11-11 07:44:25,988 INFO DagEngine:539 - USER[-] GROUP[-] TOKEN[-] APP[-]
JOB[-] ACTION[-] User hadoop killed the WF job
0000001-131111070843260-oozie-hado-W
The content of this unique workflow is as follows:
<workflow-app xmlns='uri:oozie:workflow:0.1' name='customtest'> <start
to='detectdir'/> <action name='detectdir'> <detect
xmlns="uri:custom:detect-action:0.1"> <dirname>aaa</dirname> </detect>
<ok to="end"/> <error to="fail"/> </action> <end name='end'/> <kill
name="fail"> <message>Workflow failed, error
message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill></workflow-app>The detectdir node is a custom action node,implemented by
myself.its logic is just to do a while circulation to wait for the dir
exists,like this:while(!dirname.exists()) {}so if the dir does not exist
and I submit this unique workflow and its copies 3 or more ,any other workflow
can not run.But if one or two ,they can work normally.They wait for the dir and
do not influence other workflows. here are some relevant logs,please see if
it can help：2013-11-11 08:51:38,938 INFO DagEngine:539 - USER[-] GROUP[-]
TOKEN[-] APP[-] JOB[-] ACTION[-] User hadoop killed the WFjob
0000005-131111082227563-oozie-hado-W2013-11-11 08:51:47,523 INFO DagEngine:539
- USER[-] GROUP[-] TOKEN[-] APP[-] JOB[-] ACTION[-] User hadoop killed the
WFjob 0000001-131111082227563-oozie-hado-W2013-11-11 08:52:11,742 WARN
ActionStartXCommand:542 - USER[root] GROUP[-] TOKEN[] APP['customtest']
JOB[0000001-131111082227563-oozie-hado-W]
ACTION[0000001-131111082227563-oozie-hado-W@simpleEmail]
[***0000001-131111082227563-oozie-hado-W@simpleEmail***]Action
status=DONE2013-11-11 08:52:11,743 WARN ActionStartXCommand:542 - USER[root]
GROUP[-] TOKEN[] APP['customtest'] JOB[0000001-131111082227563-oozie-hado-W]
ACTION[0000001-131111082227563-oozie-hado-W@simpleEmail]
[***0000001-131111082227563-oozie-hado-W@simpleEmail***]Action updated in
DB!2013-11-11 08:52:11,779 WARN ActionStartXCommand:542 - USER[root] GROUP[-]
TOKEN[] APP['customtest'3] JOB[0000005-131111082227563-oozie-hado-W]
ACTION[0000005-131111082227563-oozie-hado-W@simpleEmail] E0816: Action
pending=[false], status=[KILLED]. Skipping ActionStart Execution, Error Code:
E08162013-11-11 08:52:11,780 INFO KillXCommand:539 - USER[root] GROUP[-]
TOKEN[] APP['customtest'] JOB[0000001-131111082227563-oozie-hado-W] ACTION[-]
STARTED WorkflowKillXCommand for
jobId=0000001-131111082227563-oozie-hado-W2013-11-11 08:52:11,818 WARN
CoordActionUpdateXCommand:542 - USER[root] GROUP[-] TOKEN[] APP['customtest']
JOB[0000001-131111082227563-oozie-hado-W] ACTION[-] E1100: Command precondition
does not hold before execution, [, coord action is null], Error Code:
E11002013-11-11 08:52:11,818 INFO KillXCommand:539 - USER[root] GROUP[-]
TOKEN[] APP['customtest'] JOB[0000001-131111082227563-oozie-hado-W] ACTION[-]
ENDED WorkflowKillXCommand for
jobId=0000001-131111082227563-oozie-hado-W2013-11-11 08:52:11,824 WARN
ActionEndXCommand:542 - USER[root] GROUP[-] TOKEN[] APP['customtest']
JOB[0000001-13111108222 Waiting for your help sincerely!
Best Wishes!
HENRY
From: Mohammad Islam
Date: 2013-11-11 11:52
To: [email protected]; renguihe
Subject: Re: Problems with Custom Action Nodes
Hi Henry,
Can you please share the relevant contents of oozie.log and yous workflow.xml?
These will help others to address the problem.
Regards,
Mohammad
On Saturday, November 9, 2013 7:13 PM, renguihe <[email protected]> wrote:
Hi,
I try to use Custom Action Nodes.
The workflow can be submitted but hangs on "PREP" status.
I wonder which step is wrong.
I followed the steps according this article .
http://www.infoq.com/articles/ExtendingOozie
I put a jar containing EmailActionExecutor class and emailAction.xsd in the
oozie server lib.
I wonder how xsd knows to use the EmailActionExecutor class?
There is no expicit declaration in xsd.
thx!
HENRY
