Stack overflow on Generic Controller when IfController is used #2456

asfimport · 2011-01-19T10:07:05Z

Luciana Moreira (Bug 50618):
After upgrading to the newest version of JMeter on SVN (revision 1059949), I noticed that I am getting a stack overflow in the following method GenericController.reInitializeSubController().

/**
 * Called to re-initialize a index of controller's elements (https://github.com/apache/jmeter/issues/2405)
 * 
 * @return Sampler
 */
protected Sampler reInitializeSubController() {
    Sampler returnValue = null;
    try {
        TestElement currentElement = getCurrentElement();
        if (currentElement != null) {
            if (currentElement instanceof Sampler) {
                returnValue = nextIsASampler((Sampler) currentElement);
            } else { // must be a controller
                returnValue = nextIsAController((Controller) currentElement);
                reInitializeSubController();
            }
        }
    } catch (NextIsNullException e) {
    }
    return returnValue;
}

The problem happens in this else statement:
} else { // must be a controller
     returnValue = nextIsAController((Controller) currentElement);
     reInitializeSubController();
}

For some reason when I execute my script I get to this else statement, and from there on, I am not able to leave it any longer. I have attached my ThreadDump in hope that it helps you somehow.
 Any suggestions?
Luciana

Votes in Bugzilla: 2
OS: All

Duplicates:

The text was updated successfully, but these errors were encountered:

asfimport · 2011-01-19T10:09:05Z

Luciana Moreira (migrated from Bugzilla):
Created attachment threaddump.tdump: ThreadDump with the stack overflow

asfimport · 2011-01-19T12:20:46Z

Luciana Moreira (migrated from Bugzilla):
After searching a bit more I found out that this bug was introduced when fixing bug
#2405.

In the class IfController.java a new line was added by milamber on revision 10215112:
super.reInitializeSubController(); // #2405 - reinitialize current index element for all sub controller

If I comment this out, everything works out like a charm. I do not know however, what is the solution for Bug50032

asfimport · 2011-01-19T15:13:11Z

@milamberspace (migrated from Bugzilla):
Can you provide a test case for this stack overflow?
Thanks

asfimport · 2011-01-20T07:31:50Z

Luciana Moreira (migrated from Bugzilla):
Sure. I have attached a simple test case that get's into this condition.

asfimport · 2011-01-20T07:32:33Z

Luciana Moreira (migrated from Bugzilla):
Created attachment Test.jmx: Simple testcase to generate a stack overflow

Test.jmx

<?xml version="1.0" encoding="UTF-8"?>
<jmeterTestPlan version="1.2" properties="2.1">
  <hashTree>
    <TestPlan guiclass="TestPlanGui" testclass="TestPlan" testname="Test Plan" enabled="true">
      <stringProp name="TestPlan.comments"></stringProp>
      <boolProp name="TestPlan.functional_mode">false</boolProp>
      <boolProp name="TestPlan.serialize_threadgroups">false</boolProp>
      <elementProp name="TestPlan.user_defined_variables" elementType="Arguments" guiclass="ArgumentsPanel" testclass="Arguments" testname="User Defined Variables" enabled="true">
        <collectionProp name="Arguments.arguments"/>
      </elementProp>
      <stringProp name="TestPlan.user_define_classpath"></stringProp>
    </TestPlan>
    <hashTree>
      <ThreadGroup guiclass="ThreadGroupGui" testclass="ThreadGroup" testname="Test" enabled="true">
        <elementProp name="ThreadGroup.main_controller" elementType="LoopController" guiclass="LoopControlPanel" testclass="LoopController" testname="Loop Controller" enabled="true">
          <boolProp name="LoopController.continue_forever">false</boolProp>
          <intProp name="LoopController.loops">-1</intProp>
        </elementProp>
        <stringProp name="ThreadGroup.num_threads">1</stringProp>
        <stringProp name="ThreadGroup.ramp_time">1</stringProp>
        <longProp name="ThreadGroup.start_time">1216276358000</longProp>
        <longProp name="ThreadGroup.end_time">1216276358000</longProp>
        <boolProp name="ThreadGroup.scheduler">false</boolProp>
        <stringProp name="ThreadGroup.on_sample_error">continue</stringProp>
        <stringProp name="ThreadGroup.duration"></stringProp>
        <stringProp name="ThreadGroup.delay"></stringProp>
      </ThreadGroup>
      <hashTree>
        <GenericController guiclass="LogicControllerGui" testclass="GenericController" testname="Simple Controller" enabled="true"/>
        <hashTree>
          <IfController guiclass="IfControllerPanel" testclass="IfController" testname="If Controller" enabled="true">
            <stringProp name="IfController.condition">${accept_request}==1</stringProp>
            <boolProp name="IfController.evaluateAll">false</boolProp>
          </IfController>
          <hashTree>
            <GenericController guiclass="LogicControllerGui" testclass="GenericController" testname="Simple Controller" enabled="true"/>
            <hashTree>
              <BeanShellSampler guiclass="BeanShellSamplerGui" testclass="BeanShellSampler" testname="PrintOut" enabled="true">
                <stringProp name="BeanShellSampler.query">print(&quot;Testing stack overflow.&quot;);</stringProp>
                <stringProp name="BeanShellSampler.filename"></stringProp>
                <stringProp name="BeanShellSampler.parameters"></stringProp>
                <boolProp name="BeanShellSampler.resetInterpreter">false</boolProp>
              </BeanShellSampler>
              <hashTree/>
            </hashTree>
          </hashTree>
          <ResultCollector guiclass="ViewResultsFullVisualizer" testclass="ResultCollector" testname="View Results Tree" enabled="true">
            <boolProp name="ResultCollector.error_logging">false</boolProp>
            <objProp>
              <name>saveConfig</name>
              <value class="SampleSaveConfiguration">
                <time>true</time>
                <latency>true</latency>
                <timestamp>true</timestamp>
                <success>true</success>
                <label>true</label>
                <code>true</code>
                <message>true</message>
                <threadName>true</threadName>
                <dataType>true</dataType>
                <encoding>true</encoding>
                <assertions>true</assertions>
                <subresults>true</subresults>
                <responseData>true</responseData>
                <samplerData>true</samplerData>
                <xml>true</xml>
                <fieldNames>true</fieldNames>
                <responseHeaders>true</responseHeaders>
                <requestHeaders>true</requestHeaders>
                <responseDataOnError>false</responseDataOnError>
                <saveAssertionResultsFailureMessage>true</saveAssertionResultsFailureMessage>
                <assertionsResultsToSave>0</assertionsResultsToSave>
                <bytes>true</bytes>
                <url>true</url>
                <fileName>true</fileName>
                <hostname>true</hostname>
                <threadCounts>true</threadCounts>
                <sampleCount>true</sampleCount>
                <idleTime>true</idleTime>
              </value>
            </objProp>
            <stringProp name="filename"></stringProp>
          </ResultCollector>
          <hashTree/>
        </hashTree>
      </hashTree>
    </hashTree>
  </hashTree>
</jmeterTestPlan>

asfimport · 2011-01-20T20:34:02Z

@milamberspace (migrated from Bugzilla):
This StackOverflowError isn't come with patch of #2405.
With your simple testcase, you can have this SOE with the current JMeter 2.4 r961953.

This SOE comes because the IF Controller always returns false (${accept_request} isn't defines in test plan) and thus there haven't any sampler executed.

Your test case is a special case...
If you add a simple sampler (like java sampler) outside the IF Controller (before or after), the SOE is avoided.

asfimport · 2011-01-21T03:45:00Z

Luciana Moreira (migrated from Bugzilla):
I simplified my test case to only point to this bug and ended up cutting important parts it would seem...
I have created another test case which has an Include Controller and in there we have a Sampler. In this test case we also get the stack overflow.

In any case, I don't see why we would always need to add a sampler after a If Controller. Can't we have as a last task of a thread group an action executed by a Sampler inside an If Controller?
Using a big CSV file I can easily imagine many scenarios where this would be needed. According to a variable in the CSV file we do this additional step or not.

asfimport · 2011-01-21T03:45:54Z

Luciana Moreira (migrated from Bugzilla):
Created attachment Test.jmx: File with Include Controller and Sampler

asfimport · 2011-01-21T03:46:28Z

Luciana Moreira (migrated from Bugzilla):
Created attachment Include.jmx: A simpler task for a include controller

Include.jmx

<?xml version="1.0" encoding="UTF-8"?>
<jmeterTestPlan version="1.2" properties="2.1">
  <hashTree>
    <GenericController guiclass="LogicControllerGui" testclass="GenericController" testname="Simple Controller" enabled="true"/>
    <hashTree>
      <BeanShellSampler guiclass="BeanShellSamplerGui" testclass="BeanShellSampler" testname="Print Pop Test Number" enabled="true">
        <stringProp name="BeanShellSampler.query">print(&quot;Testing Include Controller&quot;));</stringProp>
        <stringProp name="BeanShellSampler.filename"></stringProp>
        <stringProp name="BeanShellSampler.parameters"></stringProp>
        <boolProp name="BeanShellSampler.resetInterpreter">false</boolProp>
      </BeanShellSampler>
      <hashTree/>
    </hashTree>
  </hashTree>
</jmeterTestPlan>

asfimport · 2011-01-21T05:09:06Z

Luciana Moreira (migrated from Bugzilla):
Created attachment Test.jmx: Excluded CSV and user defined variables

Test.jmx

<?xml version="1.0" encoding="UTF-8"?>
<jmeterTestPlan version="1.2" properties="2.1">
  <hashTree>
    <TestPlan guiclass="TestPlanGui" testclass="TestPlan" testname="Test Plan" enabled="true">
      <stringProp name="TestPlan.comments"></stringProp>
      <boolProp name="TestPlan.functional_mode">false</boolProp>
      <boolProp name="TestPlan.serialize_threadgroups">false</boolProp>
      <elementProp name="TestPlan.user_defined_variables" elementType="Arguments" guiclass="ArgumentsPanel" testclass="Arguments" testname="User Defined Variables" enabled="true">
        <collectionProp name="Arguments.arguments"/>
      </elementProp>
      <stringProp name="TestPlan.user_define_classpath"></stringProp>
    </TestPlan>
    <hashTree>
      <ThreadGroup guiclass="ThreadGroupGui" testclass="ThreadGroup" testname="Test" enabled="true">
        <elementProp name="ThreadGroup.main_controller" elementType="LoopController" guiclass="LoopControlPanel" testclass="LoopController" testname="Loop Controller" enabled="true">
          <boolProp name="LoopController.continue_forever">false</boolProp>
          <intProp name="LoopController.loops">-1</intProp>
        </elementProp>
        <stringProp name="ThreadGroup.num_threads">1</stringProp>
        <stringProp name="ThreadGroup.ramp_time">1</stringProp>
        <longProp name="ThreadGroup.start_time">1216276358000</longProp>
        <longProp name="ThreadGroup.end_time">1216276358000</longProp>
        <boolProp name="ThreadGroup.scheduler">false</boolProp>
        <stringProp name="ThreadGroup.on_sample_error">continue</stringProp>
        <stringProp name="ThreadGroup.duration"></stringProp>
        <stringProp name="ThreadGroup.delay"></stringProp>
      </ThreadGroup>
      <hashTree>
        <GenericController guiclass="LogicControllerGui" testclass="GenericController" testname="Simple Controller" enabled="true"/>
        <hashTree>
          <BeanShellSampler guiclass="BeanShellSamplerGui" testclass="BeanShellSampler" testname="PrintOut" enabled="true">
            <stringProp name="BeanShellSampler.query">print(&quot;Testing stack overflow.&quot;);</stringProp>
            <stringProp name="BeanShellSampler.filename"></stringProp>
            <stringProp name="BeanShellSampler.parameters"></stringProp>
            <boolProp name="BeanShellSampler.resetInterpreter">false</boolProp>
          </BeanShellSampler>
          <hashTree/>
          <IfController guiclass="IfControllerPanel" testclass="IfController" testname="If Controller" enabled="true">
            <stringProp name="IfController.condition">0==1</stringProp>
            <boolProp name="IfController.evaluateAll">false</boolProp>
          </IfController>
          <hashTree>
            <GenericController guiclass="LogicControllerGui" testclass="GenericController" testname="Simple Controller" enabled="true"/>
            <hashTree>
              <BeanShellSampler guiclass="BeanShellSamplerGui" testclass="BeanShellSampler" testname="PrintOut" enabled="true">
                <stringProp name="BeanShellSampler.query">print(&quot;Testing stack overflow.&quot;);</stringProp>
                <stringProp name="BeanShellSampler.filename"></stringProp>
                <stringProp name="BeanShellSampler.parameters"></stringProp>
                <boolProp name="BeanShellSampler.resetInterpreter">false</boolProp>
              </BeanShellSampler>
              <hashTree/>
            </hashTree>
          </hashTree>
          <IncludeController guiclass="IncludeControllerGui" testclass="IncludeController" testname="Include Controller" enabled="true">
            <stringProp name="IncludeController.includepath">/home/moreira/workspace/HudsonTests/trunk/jmeter/scripts/PopMail.jmx</stringProp>
          </IncludeController>
          <hashTree/>
          <ResultCollector guiclass="ViewResultsFullVisualizer" testclass="ResultCollector" testname="View Results Tree" enabled="true">
            <boolProp name="ResultCollector.error_logging">false</boolProp>
            <objProp>
              <name>saveConfig</name>
              <value class="SampleSaveConfiguration">
                <time>true</time>
                <latency>true</latency>
                <timestamp>true</timestamp>
                <success>true</success>
                <label>true</label>
                <code>true</code>
                <message>true</message>
                <threadName>true</threadName>
                <dataType>true</dataType>
                <encoding>true</encoding>
                <assertions>true</assertions>
                <subresults>true</subresults>
                <responseData>true</responseData>
                <samplerData>true</samplerData>
                <xml>true</xml>
                <fieldNames>true</fieldNames>
                <responseHeaders>true</responseHeaders>
                <requestHeaders>true</requestHeaders>
                <responseDataOnError>false</responseDataOnError>
                <saveAssertionResultsFailureMessage>true</saveAssertionResultsFailureMessage>
                <assertionsResultsToSave>0</assertionsResultsToSave>
                <bytes>true</bytes>
                <url>true</url>
                <fileName>true</fileName>
                <hostname>true</hostname>
                <threadCounts>true</threadCounts>
                <sampleCount>true</sampleCount>
                <idleTime>true</idleTime>
              </value>
            </objProp>
            <stringProp name="filename"></stringProp>
          </ResultCollector>
          <hashTree/>
        </hashTree>
      </hashTree>
    </hashTree>
  </hashTree>
</jmeterTestPlan>

asfimport · 2011-01-21T09:52:32Z

Mark Thomas (migrated from Bugzilla):
The content of attachment 26527 has been deleted by
Mark Thomas <markt@apache.org>
who provided the following reason:

Inadvertent disclosure of sensitive information

The token used to delete this attachment was generated at 2011-01-21 14:51:02 GMT.

asfimport · 2011-01-23T14:23:46Z

@milamberspace (migrated from Bugzilla):
This bug seems very old. It's reproducible since at least version 2.3.4.
It seems to link to script with no single sampler, only a Simple Controller/If Controller.
I need some time for better understand the root cause. (but I don't have the time from tonight to next month...)
I attached a simple script to reproduce the StackOverFlowError

asfimport · 2011-01-23T14:24:43Z

@milamberspace (migrated from Bugzilla):
Created attachment TestSimple.jmx: Simple test to reproduce error

asfimport · 2011-02-03T05:41:26Z

Luciana Moreira (migrated from Bugzilla):
I just saw that line 173 in the IfController changed from super.reInitializeSubController() to reInitializeSubController().

Even though this is an advance toward a solution in this bug, it does not yet solve the problem. Unfortunatelly the IfController does not implement this method and we once again end up in the GenericController with the stack overflow.

This is unfortunate since I have several tests that used to work perfectly and now are out of operation. Is there any plan on fixing this?

asfimport · 2011-02-06T09:44:52Z

@milamberspace (migrated from Bugzilla):
(In reply to comment 14)

I just saw that line 173 in the IfController changed from
super.reInitializeSubController() to reInitializeSubController().

Even though this is an advance toward a solution in this bug, it does not yet
solve the problem.

No, unfortunately, it is not an advance toward a solution.

Unfortunatelly the IfController does not implement this
method and we once again end up in the GenericController with the stack
overflow.

This is unfortunate since I have several tests that used to work perfectly and
now are out of operation. Is there any plan on fixing this?

What is the version of JMeter that works with your tests ?

I will works to fix this bug, I just need some time to find the time to do this work.

asfimport · 2011-02-08T09:38:03Z

Luciana Moreira (migrated from Bugzilla):
Hello Milamber,

I just synchronized to revision 1021511 on SVN and run my tests without any issues.

Hope this helps.
Luciana

asfimport · 2011-02-12T18:21:39Z

@milamberspace (migrated from Bugzilla):
With revision 1021511, I can reproduce stack overflow with the simple test attached to this bug.

asfimport · 2011-02-16T10:33:03Z

Luciana Moreira (migrated from Bugzilla):
Hello Milamber,

I just run your test and it indeed remains with the problem. Nevertheless, in this revision my test case runs with no problem.

It looks to me that our test cases show 2 different faces of the same problem. Mine had a stack overflow on the IfController, and yours seem to generate this problem in the LoopController.

In revision 1021511 the problem with the IfController was not yet there, and my test cases did not fall into the category that trigger the issue with the LoopController.

asfimport · 2011-08-30T23:58:18Z

Christian Seeberger (migrated from Bugzilla):
not only the "if" controller is concerned by this bug, but also other controllers following a controller (e.g. while controller).
I found the following workaround: Just ad a "Test Action" Sampler between the two concerned controllers and configure it with a pause duration of 1ms.

asfimport · 2011-09-09T18:18:47Z

@milamberspace (migrated from Bugzilla):

This bug comes from a infinite recursive method. JMeter uses next() method to come down the element tree.
If test plan with loops forever (or >2000 loops) contains only a If Controller with children elements, and the IF condition become "false", the StackOverFlowError will start.

The If Controller uses the next() method to provide own feature, if the condition is true, then If Controller next() call super.next() and JMeter calls the first child, if the condition if false, then If Controller next() return null (meaning for JMeter: end of controller).
The issue is providing by GenericController and nextIsController() method which come down in tree recursively, and without stop (i.e. return to JMeterThread) if always null is return by a controller.
With a unique If Controller which start to return always null, the StackOverFlow is inevitable.

A workaround is add a simple sampler like Debug, Test action on the same level of If Controller.

This first patch is to catch the StackOverFlowError and force JMeter to return on JMeterThread (and ignore the SOF error). It's works with the simple test case (TestSimple.jmx) (first iteration return true and the next return always false)

But this patch don't work if the If Controller return always false (from iteration 1 to infinite) (to test it, change If condition to
"1" == "0"

@sebb: I would like commit this first patch because it's resolving some case. Are you ok? or any opinion?
Perhaps, we add an "known issue" in documentation?

@ALL: I spent some hours (day) to find a way to correct this bug. I'm interested by your feedback to help to fix this bug.

Thanks

Created attachment bug_50618_patch1.patch: First patch to catch StackOverFlowError

bug_50618_patch1.patch

Index: src/core/org/apache/jmeter/control/GenericController.java
===================================================================
--- src/core/org/apache/jmeter/control/GenericController.java	(revision 1166154)
+++ src/core/org/apache/jmeter/control/GenericController.java	(working copy)
@@ -181,7 +181,14 @@
      * @throws NextIsNullException
      */
     protected Sampler nextIsAController(Controller controller) throws NextIsNullException {
-        Sampler sampler = controller.next();
+        Sampler sampler = null;
+        try {
+            sampler = controller.next();
+        } catch (StackOverflowError soe) {
+            // Special case: see Bug 50618
+            log.info("StackOverflowError detected");
+            throw new NextIsNullException();
+        }
         if (sampler == null) {
             currentReturnedNull(controller);
             sampler = next();

asfimport · 2011-09-09T18:36:23Z

Sebb (migrated from Bugzilla):
(In reply to comment 22)

Created attachment 27478 [details]
First patch to catch StackOverFlowError

Patch looks good, but I would use log.warn rather than log.info.

asfimport · 2011-09-12T07:18:15Z

@pmouawad (migrated from Bugzilla):
Hello,
For me issue comes from the fact that code in else statement of reInitializeSubController does not change any state, particularly nextIsAController:
nextIsAController((Controller) currentElement);
reInitializeSubController();

so starting from there reInitializeSubController will recurse infinitely.

I attach the same scenario I attached on 50032.

Regards
Philippe

asfimport · 2011-09-12T07:29:59Z

@pmouawad (migrated from Bugzilla):
(In reply to comment 24)

Hello,
For me issue comes from the fact that code in else statement of
reInitializeSubController does not change any state, particularly
nextIsAController:
nextIsAController((Controller) currentElement);
reInitializeSubController();

so starting from there reInitializeSubController will recurse infinitely.

I attach the same scenario I attached on 50032.

Regards
Philippe

Please ignore this comment, my analysis is wrong.

asfimport · 2011-09-12T07:51:34Z

@pmouawad (migrated from Bugzilla):
Hello,
Patch may fix the issue but don't you think it can introduce a big performance cost when it occurs ?

Milamber can you explain a little more the reInitializeSubController that fixed the 50032, maybe it could help us help you.

Regards
Philippe

asfimport · 2011-09-12T08:02:29Z

@pmouawad (migrated from Bugzilla):
Again,
And a little complement to what I said about catching StackOverflowError:

An Error is a subclass of Throwable that indicates serious problems that a reasonable application should not try to catch. Most such errors are abnormal conditions. The ThreadDeath error, though a "normal" condition, is also a subclass of Error because most applications should not try to catch it. A method is not required to declare in its throws clause any subclasses of Error that might be thrown during the execution of the method but not caught, since these errors are abnormal conditions that should never occur.

Regards
Philippe

asfimport · 2011-09-12T09:52:12Z

@pmouawad (migrated from Bugzilla):
Hello,
Here is my fix proposition.
I just test that nextIsAController does not return null before executing reInitializeSubController a new time.

I tested with my JMX and the JMX in case 50032 and it seems ok.

But once again I think there is still an issue in #2405 fix because it gives wrong results regarding TransactionController inside IfCOntroller, they will be mentionnned as sampled even if condition is false due to what reInitializeSubController does.

Regards
Philippe Mouawad

Created attachment BUG_50618.patch: Fix the StackOverFlow

BUG_50618.patch

Index: src/core/org/apache/jmeter/control/GenericController.java
===================================================================
--- src/core/org/apache/jmeter/control/GenericController.java	(revision 1166194)
+++ src/core/org/apache/jmeter/control/GenericController.java	(working copy)
@@ -226,8 +226,11 @@
                 if (currentElement instanceof Sampler) {
                     nextIsASampler((Sampler) currentElement);
                 } else { // must be a controller
-                    nextIsAController((Controller) currentElement);
-                    reInitializeSubController();
+                    Sampler s = nextIsAController((Controller) currentElement);
+                    if(s != null)
+                    {
+                        reInitializeSubController();                        
+                    }
                 }
             }
         } catch (NextIsNullException e) {

asfimport · 2011-09-13T23:42:54Z

@milamberspace (migrated from Bugzilla):
(In reply to comment 28)

Created attachment 27488 [details]
Fix the StackOverFlow

This patch fixes only infinite loop (without stack overflow error on jmeter.log). This is a regression from #2405.
See: #2405 (comment)

SOE already exists with (for example) TestSimple.jmx https://issues.apache.org/bugzilla/attachment.cgi?id=26539

asfimport · 2011-09-14T00:02:12Z

@milamberspace (migrated from Bugzilla):
(In reply to comment 27)

Again,
And a little complement to what I said about catching StackOverflowError:

An Error is a subclass of Throwable that indicates serious problems that a
reasonable application should not try to catch. Most such errors are abnormal
conditions. The ThreadDeath error, though a "normal" condition, is also a
subclass of Error because most applications should not try to catch it. A
method is not required to declare in its throws clause any subclasses of Error
that might be thrown during the execution of the method but not caught, since
these errors are abnormal conditions that should never occur.

Regards
Philippe

We are in a special case: infinite recursive loop without exit condition when If controller begin to return null (because the condition return false). (Like a branch on a JMeter tree without end - very big branch)
It's seems very (very) difficult to solve, a way is to introduce a limit on each call (in recursive method), and test this limit to stop loop. This way is not good for JMeter (introduce a limit for If Controller)
With Java, the try catch block is a good way to catch this 'technical' error (the stack is full), and to force JMeter to return on root tree, and start a new turn to fill the stack, etc.

A problem still the test case when If Controller always return false (from first loop). JMeter doesn't count iteration, and start a long running without end...

asfimport · 2011-09-14T00:10:07Z

@milamberspace (migrated from Bugzilla):
(In reply to comment 22)

Created attachment 27478 [details]
First patch to catch StackOverFlowError

URL: http://svn.apache.org/viewvc?rev=1170396&view=rev
Log:
If Controller - Catches a StackOverflowError when a condition returns always false (after at least one iteration with return true) See #2456

Modified:
jakarta/jmeter/trunk/src/core/org/apache/jmeter/control/GenericController.java
jakarta/jmeter/trunk/xdocs/changes.xml

asfimport · 2011-09-14T16:44:52Z

@milamberspace (migrated from Bugzilla):
URL: http://svn.apache.org/viewvc?rev=1170707&view=rev
Log:
Add two knows issues : #2405 and #2456

Modified:
jakarta/jmeter/trunk/xdocs/changes.xml

asfimport · 2011-12-28T21:47:08Z

@pmouawad (migrated from Bugzilla):
(In reply to comment 30)

(In reply to comment 27)
> Again,
> And a little complement to what I said about catching StackOverflowError:
>
> An Error is a subclass of Throwable that indicates serious problems that a
> reasonable application should not try to catch. Most such errors are abnormal
> conditions. The ThreadDeath error, though a "normal" condition, is also a
> subclass of Error because most applications should not try to catch it. A
> method is not required to declare in its throws clause any subclasses of Error
> that might be thrown during the execution of the method but not caught, since
> these errors are abnormal conditions that should never occur.
>
>
> Regards
> Philippe

We are in a special case: infinite recursive loop without exit condition when
If controller begin to return null (because the condition return false). (Like
a branch on a JMeter tree without end - very big branch)
It's seems very (very) difficult to solve, a way is to introduce a limit on
each call (in recursive method), and test this limit to stop loop. This way is
not good for JMeter (introduce a limit for If Controller)
With Java, the try catch block is a good way to catch this 'technical' error
(the stack is full), and to force JMeter to return on root tree, and start a
new turn to fill the stack, etc.

A problem still the test case when If Controller always return false (from
first loop). JMeter doesn't count iteration, and start a long running without
end...

Issue seems to be fixed even with always false condition.

asfimport · 2012-01-22T13:13:45Z

@pmouawad (migrated from Bugzilla):
As discussed on mailing list, closing issue as this particular one is fixed.
Opened #2708 for counts number that don't increase in Listeners.

asfimport closed this as completed Jan 22, 2012

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Stack overflow on Generic Controller when IfController is used #2456

Stack overflow on Generic Controller when IfController is used #2456

asfimport commented Jan 19, 2011

asfimport commented Jan 19, 2011

asfimport commented Jan 19, 2011

asfimport commented Jan 19, 2011

asfimport commented Jan 20, 2011

asfimport commented Jan 20, 2011

asfimport commented Jan 20, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 23, 2011

asfimport commented Jan 23, 2011

asfimport commented Feb 3, 2011

asfimport commented Feb 6, 2011

asfimport commented Feb 8, 2011

asfimport commented Feb 12, 2011

asfimport commented Feb 16, 2011

asfimport commented Aug 30, 2011

asfimport commented Sep 9, 2011

asfimport commented Sep 9, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 13, 2011

asfimport commented Sep 14, 2011

asfimport commented Sep 14, 2011

asfimport commented Sep 14, 2011

asfimport commented Dec 28, 2011

asfimport commented Jan 22, 2012

Stack overflow on Generic Controller when IfController is used #2456

Stack overflow on Generic Controller when IfController is used #2456

Comments

asfimport commented Jan 19, 2011

asfimport commented Jan 19, 2011

asfimport commented Jan 19, 2011

asfimport commented Jan 19, 2011

asfimport commented Jan 20, 2011

asfimport commented Jan 20, 2011

asfimport commented Jan 20, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 21, 2011

asfimport commented Jan 23, 2011

asfimport commented Jan 23, 2011

asfimport commented Feb 3, 2011

asfimport commented Feb 6, 2011

asfimport commented Feb 8, 2011

asfimport commented Feb 12, 2011

asfimport commented Feb 16, 2011

asfimport commented Aug 30, 2011

asfimport commented Sep 9, 2011

asfimport commented Sep 9, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 12, 2011

asfimport commented Sep 13, 2011

asfimport commented Sep 14, 2011

asfimport commented Sep 14, 2011

asfimport commented Sep 14, 2011

asfimport commented Dec 28, 2011

asfimport commented Jan 22, 2012