You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by Ole Solberg <Ol...@Sun.COM> on 2009/04/03 13:59:27 UTC

Preliminary results from testing the 10.5.1.0 RC

The following failures were seen:


DERBY-2478 (CLOSED!?)

DERBY-2687 (Intermittent)

DERBY-2447 (Intermittent)

DERBY-4097 (Intermittent)

DERBY-3819 (Unexpected query plan... / Intermittent?)

DERBY-4118 (on one single os - 'Unable to analyze class path: access
denied (java.util.PropertyPermission java.class.path read)'
    jars in different order in classpath?

Replication test noise! (Sorry, *I* have a job to do here ....)

testInertTime(...TimeHandlingTest)junit.framework.AssertionFailedError:
expected:<2> but was:<3>


---------------------------------------------------
jvm1.6-64
http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/Limited/ :

1 X DERBY-2478 on "Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits" - CLOSED?

8 X DERBY-3819 (1 X 8 platforms)

6 X DERBY-4118 on "Red Hat Enterprise Linux Server release 5.1 (Tikanga)
64bits"

8+ X replication test "noise" ....


---------------------------------------------------
jvm1.6
http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6/testing/Limited/ :

1 X DERBY-2687 on "Solaris 10 5/08 s10x_u5wos_10 X86 64bits" (1 of 2
Solaris zones running concurrently.)

6 X DERBY-4118 on "Red Hat Enterprise Linux Server release 5.1 (Tikanga)
64bits"

8 X replication test "noise" ....


---------------------------------------------------
jvm1.5
http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.5/testing/Limited/ :
1 X DERBY-2447 on "Red Hat Enterprise Linux Server release 5.1 (Tikanga)
32bits"

1 X DERBY-4097 on "Solaris 10 8/07 s10x_u4wos_12b X86 64bits" (1 of 2
Solaris zones running concurrently.)

6 X DERBY-4118 on "Red Hat Enterprise Linux Server release 5.1 (Tikanga)
64bits"

4 X
testInertTime(...TimeHandlingTest)junit.framework.AssertionFailedError:
expected:<2> but was:<3>


2 X replication test "noise" ....


---------------------------------------------------
jvm1.4
http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.4/testing/Limited/ :

6 X DERBY-4118 on "Red Hat Enterprise Linux Server release 5.1 (Tikanga)
64bits"

6 X
testInertTime(...TimeHandlingTest)junit.framework.AssertionFailedError:
expected:<2> but was:<3>

2 X replication test "noise" ....


---------------------------------------------------
Some platforms need to be rerun due to "hangs", probably caused by env.
trouble....


-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/06/09 22:47, Myrna van Lunteren wrote:
> Hi,
> 
> Is it possible to fill out the details on the platform testing page:
> http://wiki.apache.org/db-derby/TenFiveOnePlatformTesting
> ?
> 
> Thx,
> Myrna

I will, tomorrow morning ;-)


-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Myrna van Lunteren <m....@gmail.com>.
Hi,

Is it possible to fill out the details on the platform testing page:
http://wiki.apache.org/db-derby/TenFiveOnePlatformTesting
?

Thx,
Myrna

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/07/09 10:01, Ole Solberg wrote:
> On 04/07/2009 09:57 AM Ole Solberg wrote:
>> On 04/07/2009 09:38 AM Knut Anders Hatlen wrote:
>>> Ole Solberg <Ol...@Sun.COM> writes:
>>>
>>>> I did another run on sparcN+1 aka Solaris Express Community Edition
>>>> snv_95 SPARC.
>>>> (And on sparc aka Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits, which is
>>>> not yet finished.)
>>>>
>>>> The tests do NOT hang but takes more than 10 hrs on these platforms,
>>>> when our test environment just kicks it out!
>>> Could it be this bug? http://bugs.sun.com/view_bug.do?bug_id=6471009
>>>
>>> I seem to recall that it used to make suites.All take a very long time
>>> on older releases of Java RE 6. It was only seen on the server VM, I
>>> think, but the 64-bit JVM only supports the server VM, so you can't work
>>> around the problem by using the client VM. Could you try with a newer
>>> JVM and see if the problem is still there?
>>>
>> Yes, I have started on that.
>> Will probably have our lab-people install the newer jvm though, for
>> general use in our testing.
>>
>>
> Actually, we have it, I just need to configure for it....
> 
> 

Reran OK (with the usual DERBY-3819) on

sparc aka Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits:
----------------------------------------------------
Time: 13,247.624
There was 1 failure:
1) 
test_predicatePushdown(org.apache.derbyTesting.functionTests.tests.lang.PredicatePushdownTest)junit.framework.AssertionFailedError: 
Expected Table Scan ResultSet for T3

and
sparcN+1 aka Solaris Express Community Edition snv_95 SPARC 64bits:
-------------------------------------------------------------------
Time: 13,939.859
There was 1 failure:
1) 
test_predicatePushdown(org.apache.derbyTesting.functionTests.tests.lang.PredicatePushdownTest)junit.framework.AssertionFailedError: 
Expected Table Scan ResultSet for T3



-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/07/2009 09:57 AM Ole Solberg wrote:
> On 04/07/2009 09:38 AM Knut Anders Hatlen wrote:
>> Ole Solberg <Ol...@Sun.COM> writes:
>>
>>> I did another run on sparcN+1 aka Solaris Express Community Edition
>>> snv_95 SPARC.
>>> (And on sparc aka Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits, which is
>>> not yet finished.)
>>>
>>> The tests do NOT hang but takes more than 10 hrs on these platforms,
>>> when our test environment just kicks it out!
>> Could it be this bug? http://bugs.sun.com/view_bug.do?bug_id=6471009
>>
>> I seem to recall that it used to make suites.All take a very long time
>> on older releases of Java RE 6. It was only seen on the server VM, I
>> think, but the 64-bit JVM only supports the server VM, so you can't work
>> around the problem by using the client VM. Could you try with a newer
>> JVM and see if the problem is still there?
>>
> 
> Yes, I have started on that.
> Will probably have our lab-people install the newer jvm though, for
> general use in our testing.
> 
> 
Actually, we have it, I just need to configure for it....


-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/07/2009 09:38 AM Knut Anders Hatlen wrote:
> Ole Solberg <Ol...@Sun.COM> writes:
> 
>> I did another run on sparcN+1 aka Solaris Express Community Edition
>> snv_95 SPARC.
>> (And on sparc aka Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits, which is
>> not yet finished.)
>>
>> The tests do NOT hang but takes more than 10 hrs on these platforms,
>> when our test environment just kicks it out!
> 
> Could it be this bug? http://bugs.sun.com/view_bug.do?bug_id=6471009
> 
> I seem to recall that it used to make suites.All take a very long time
> on older releases of Java RE 6. It was only seen on the server VM, I
> think, but the 64-bit JVM only supports the server VM, so you can't work
> around the problem by using the client VM. Could you try with a newer
> JVM and see if the problem is still there?
> 

Yes, I have started on that.
Will probably have our lab-people install the newer jvm though, for
general use in our testing.


-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Knut Anders Hatlen <Kn...@Sun.COM>.
Ole Solberg <Ol...@Sun.COM> writes:

> I did another run on sparcN+1 aka Solaris Express Community Edition
> snv_95 SPARC.
> (And on sparc aka Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits, which is
> not yet finished.)
>
> The tests do NOT hang but takes more than 10 hrs on these platforms,
> when our test environment just kicks it out!

Could it be this bug? http://bugs.sun.com/view_bug.do?bug_id=6471009

I seem to recall that it used to make suites.All take a very long time
on older releases of Java RE 6. It was only seen on the server VM, I
think, but the 64-bit JVM only supports the server VM, so you can't work
around the problem by using the client VM. Could you try with a newer
JVM and see if the problem is still there?

-- 
Knut Anders

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/06/09 14:21, Knut Anders Hatlen wrote:
> Ole Solberg <Ol...@Sun.COM> writes:
> 
>> I looked a little closer at this:
>> The failure appears to be in
>> Embedded_40/PredicatePushdownTest/test_predicatePushdown.
> 
> That's probably the known instability in that test, then. (DERBY-3819)
> 
>> The test run later hangs and is aborted by the testing environement....
>>
>> The full log of the test run is here:
>> http://dbtg.thresher.com/derby/test/debug/10.5.1.0_RC_jvm1.6-64_sparc_suitesAll.zip
>>
>> This was on Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits.
>> A similar hang is also seen on Solaris Express Community Edition snv_95
>> SPARCbits.
> 
> Do you know which test is hanging? Is it possible to modify your
> environment so that it takes a thread dump before aborting the test?
> (Maybe you've already done so, but I couldn't find a thread dump in the
> zip file.)
> 

I did another run on sparcN+1 aka Solaris Express Community Edition 
snv_95 SPARC.
(And on sparc aka Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits, which is 
not yet finished.)

The tests do NOT hang but takes more than 10 hrs on these platforms, 
when our test environment just kicks it out!
I did create some thread dumps on these runs to hopefully get an idea of 
what is happening.
The log from the "sparcN+1" is at
http://dbtg.thresher.com/derby/test/debug/10.5.1.0_RC_jvm1.6-64_sparcN+1_suitesAll.zip

I will upload the result of the "sparc" test when it's ready.

I will then run the tests in the pure Apache Derby test environment 
and/or run in our test environment with a much larger limit than 10 hrs.


-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Knut Anders Hatlen <Kn...@Sun.COM>.
Ole Solberg <Ol...@Sun.COM> writes:

> I looked a little closer at this:
> The failure appears to be in
> Embedded_40/PredicatePushdownTest/test_predicatePushdown.

That's probably the known instability in that test, then. (DERBY-3819)

> The test run later hangs and is aborted by the testing environement....
>
> The full log of the test run is here:
> http://dbtg.thresher.com/derby/test/debug/10.5.1.0_RC_jvm1.6-64_sparc_suitesAll.zip
>
> This was on Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits.
> A similar hang is also seen on Solaris Express Community Edition snv_95
> SPARCbits.

Do you know which test is hanging? Is it possible to modify your
environment so that it takes a thread dump before aborting the test?
(Maybe you've already done so, but I couldn't find a thread dump in the
zip file.)

-- 
Knut Anders

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/05/2009 11:47 PM Knut Anders Hatlen wrote:
> Ole Solberg <Ol...@Sun.COM> writes:
> 
>> On 04/03/09 14:27, Knut Anders Hatlen wrote:
>>> Ole Solberg <Ol...@Sun.COM> writes:
>>>
>>>> The following failures were seen:
>>> Thanks for the report, Ole. I have one question, please see below.
>>>
>>>> DERBY-2478 (CLOSED!?)
>>> Do you have the stack trace for this error? The report seems to have
>>> been truncated:
>>> http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-suitesAll_diff.txt
>>>
>> This happened in derbyall, see:
>> http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-derbyall_diff.txt
>>
>> but the first diff is
>>
>> ********* Diff file
>> derbyall/derbynetmats/DerbyNet/derbynetmats/holdCursorIJ.diff
>> *** Start: holdCursorIJ jdk1.6.0_06 DerbyNet derbynetmats:derbynetmats
>> 2009-04-02 23:25:38 ***
>> 115 del
>> < ERROR X0X95: Operation 'DROP TABLE' cannot be performed on object
>> T1' because there is an open ResultSet dependent on that object.
>> 115a115
>>> ERROR X0X95: DB2 SQL error: SQLCODE: -1, SQLSTATE: X0X95, 
>> SQLERRMC: DROP TABLET1X0X95.S
>> 116a117
>>> ERROR 58009: Execution failed due to a distribution protocol error 
>> that caused deallocation of the conversation.  The requested command
>> encountered an unarchitected and implementation specific condition for
>> which there was no architected message.
>> 120 del
>> < ERROR X0X95: Operation 'DROP TABLE' cannot be performed on object
>> T1' because there is an open ResultSet dependent on that object.
>>
>>
>> so my automatic matching probably fooled me here...
> 
> OK. But still, the link I posted above shows a failure in suites.All,
> but it doesn't say which test. Is that log from an aborted test run?
> 

I looked a little closer at this:
The failure appears to be in
Embedded_40/PredicatePushdownTest/test_predicatePushdown.

The test run later hangs and is aborted by the testing environement....

The full log of the test run is here:
http://dbtg.thresher.com/derby/test/debug/10.5.1.0_RC_jvm1.6-64_sparc_suitesAll.zip

This was on Solaris 10 5/08 s10s_u5wos_10 SPARC 64bits.
A similar hang is also seen on Solaris Express Community Edition snv_95
SPARCbits.


-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Knut Anders Hatlen <Kn...@Sun.COM>.
Ole Solberg <Ol...@Sun.COM> writes:

> On 04/03/09 14:27, Knut Anders Hatlen wrote:
>> Ole Solberg <Ol...@Sun.COM> writes:
>>
>>> The following failures were seen:
>>
>> Thanks for the report, Ole. I have one question, please see below.
>>
>>> DERBY-2478 (CLOSED!?)
>>
>> Do you have the stack trace for this error? The report seems to have
>> been truncated:
>> http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-suitesAll_diff.txt
>>
>
> This happened in derbyall, see:
> http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-derbyall_diff.txt
>
> but the first diff is
>
> ********* Diff file
> derbyall/derbynetmats/DerbyNet/derbynetmats/holdCursorIJ.diff
> *** Start: holdCursorIJ jdk1.6.0_06 DerbyNet derbynetmats:derbynetmats
> 2009-04-02 23:25:38 ***
> 115 del
> < ERROR X0X95: Operation 'DROP TABLE' cannot be performed on object
> T1' because there is an open ResultSet dependent on that object.
> 115a115
>> ERROR X0X95: DB2 SQL error: SQLCODE: -1, SQLSTATE: X0X95, 
> SQLERRMC: DROP TABLET1X0X95.S
> 116a117
>> ERROR 58009: Execution failed due to a distribution protocol error 
> that caused deallocation of the conversation.  The requested command
> encountered an unarchitected and implementation specific condition for
> which there was no architected message.
> 120 del
> < ERROR X0X95: Operation 'DROP TABLE' cannot be performed on object
> T1' because there is an open ResultSet dependent on that object.
>
>
> so my automatic matching probably fooled me here...

OK. But still, the link I posted above shows a failure in suites.All,
but it doesn't say which test. Is that log from an aborted test run?

-- 
Knut Anders

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Ole Solberg <Ol...@Sun.COM>.
On 04/03/09 14:27, Knut Anders Hatlen wrote:
> Ole Solberg <Ol...@Sun.COM> writes:
> 
>> The following failures were seen:
> 
> Thanks for the report, Ole. I have one question, please see below.
> 
>> DERBY-2478 (CLOSED!?)
> 
> Do you have the stack trace for this error? The report seems to have
> been truncated:
> http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-suitesAll_diff.txt
> 

This happened in derbyall, see:
http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-derbyall_diff.txt

but the first diff is

********* Diff file 
derbyall/derbynetmats/DerbyNet/derbynetmats/holdCursorIJ.diff
*** Start: holdCursorIJ jdk1.6.0_06 DerbyNet derbynetmats:derbynetmats 
2009-04-02 23:25:38 ***
115 del
< ERROR X0X95: Operation 'DROP TABLE' cannot be performed on object 'T1' 
because there is an open ResultSet dependent on that object.
115a115
 > ERROR X0X95: DB2 SQL error: SQLCODE: -1, SQLSTATE: X0X95, 
SQLERRMC: DROP TABLET1X0X95.S
116a117
 > ERROR 58009: Execution failed due to a distribution protocol error 
that caused deallocation of the conversation.  The requested command 
encountered an unarchitected and implementation specific condition for 
which there was no architected message.
120 del
< ERROR X0X95: Operation 'DROP TABLE' cannot be performed on object 'T1' 
because there is an open ResultSet dependent on that object.


so my automatic matching probably fooled me here...



-- 
Ole Solberg, Database Technology Group,
Sun Microsystems, Trondheim, Norway

Re: Preliminary results from testing the 10.5.1.0 RC

Posted by Knut Anders Hatlen <Kn...@Sun.COM>.
Ole Solberg <Ol...@Sun.COM> writes:

> The following failures were seen:

Thanks for the report, Ole. I have one question, please see below.

> DERBY-2478 (CLOSED!?)

Do you have the stack trace for this error? The report seems to have
been truncated:
http://dbtg.thresher.com/derby/test/10.5.1.0_RC/jvm1.6-64/testing/testlog/sparc/757599-suitesAll_diff.txt

-- 
Knut Anders