Hi,

I already noticed that on my local machine (Thinkpad Laptop with SSD), but
I was thinking it might be my slow IO system (thanks Robert for this
running-gag). But when reviewing test times on the new Jenkins Windows
build server, this test takes very long (around 30 minutes, which is half
of the complete time when running with 1 test runner on 2 cores).

There must be something that makes this test very slow on Windows:
http://goo.gl/irDVw

The method taking so long is actually "testStressRecovery", I would like
to @Ignore it until it is fixed. It makes running Solr tests in Windows
take ages and stops me from running them at all :(

Uwe
---------------------------------------------------------------------

> Hi,
>
> I already noticed that on my local machine (Thinkpad Laptop with SSD), but
> I was thinking it might be my slow IO system (thanks Robert for this
> running-gag). But when reviewing test times on the new Jenkins Windows
> build server, this test takes very long (around 30 minutes, which is half
> of the complete time when running with 1 test runner on 2 cores).
>
> There must be something that makes this test very slow on Windows:
> http://goo.gl/irDVw
>
> The method taking so long is actually "testStressRecovery", I would like
> to @Ignore it until it is fixed. It makes running Solr tests in Windows
> take ages and stops me from running them at all :(

One thing to add: the test not always takes so long, sometoimes it is
quite fast. But if it is slow it is extremely slow on windows. Maybe a
codec issue?

Uwe

---------------------------------------------------------------------

> I would like to @Ignore it until it is fixed.

You could add an assumption in @beforeclass and ignore it on windows only?

> One thing to add: the test not always takes so long, sometoimes it is
> quite fast. But if it is slow it is extremely slow on windows. Maybe a
> codec issue?

Can see the history and repeat with the same seed? If it is repeatable
then a simple method execution profile should tell what's taking so
long.

Dawid

---------------------------------------------------------------------

>> I would like to @Ignore it until it is fixed.
>
> You could add an assumption in @beforeclass and ignore it on windows only?
>
>> One thing to add: the test not always takes so long, sometoimes it is
>> quite fast. But if it is slow it is extremely slow on windows. Maybe a
>> codec issue?
>
> Can see the history and repeat with the same seed? If it is repeatable
> then a simple method execution profile should tell what's taking so
> long.

How top get the seed? The test passed and I only see the ANT output like
listed in the above URL. I only see the master seed, does this help to
execute this one test isolated?

Sorry, I am uninformed about randomizedrunner :-)

Uwe

---------------------------------------------------------------------

> How top get the seed? The test passed and I only see the ANT output like
> listed in the above URL. I only see the master seed, does this help to
> execute this one test isolated?

The master seed is what matters. Everything else is a derivation of
that -- if you filter (narrow down to a particular method) that method
will still get an identical starting seed derived from the master.
That's also the reason the master seed is always printed by junit4 as
in here:

    [mkdir] Created dir:
C:\Jenkins\workspace\Windows-Lucene-Solr-tests-only-trunk-java7\solr\build\solr-core\test
   [junit4] <JUnit4> says aloha! Master seed: 4877EC7806E56AF3

Note that the execution here:
http://jenkins.sd-datasolutions.de/job/Windows-Lucene-Solr-tests-only-trunk-java7/11/consoleText

had errors prior to the "long" test -- maybe this is what's causing
the slowdown somehow? I'd try to re-run the same set of classes with
the same master seed, without narrowing to that particular suite; may
be a dependency issue.

> Sorry, I am uninformed about randomizedrunner :-)

Not a problem. I'll try to provide feedback; ping me directly if I
miss an e-mail or a problem (it can happen given the amount of email
junk I get).

Dawid

---------------------------------------------------------------------

OK, found out:

The test runs endless also on my local machine, what happened:

- SerialMergeScheduler
- SimpleText codec

That says everything. I have no idea why it is faster on Linux, but that
is a quite common config I get here also with other seeds.

So we should annotate that test with the SuppressCodecs() annotation. Any
ideas what codecs to suppress?

>> How top get the seed? The test passed and I only see the ANT output like
>> listed in the above URL. I only see the master seed, does this help to
>> execute this one test isolated?
>
> The master seed is what matters. Everything else is a derivation of
> that -- if you filter (narrow down to a particular method) that method
> will still get an identical starting seed derived from the master.
> That's also the reason the master seed is always printed by junit4 as
> in here:
>
>     [mkdir] Created dir:
> C:\Jenkins\workspace\Windows-Lucene-Solr-tests-only-trunk-java7\solr\build\solr-core\test
>    [junit4] <JUnit4> says aloha! Master seed: 4877EC7806E56AF3
>
> Note that the execution here:
> http://jenkins.sd-datasolutions.de/job/Windows-Lucene-Solr-tests-only-trunk-java7/11/consoleText
>
> had errors prior to the "long" test -- maybe this is what's causing
> the slowdown somehow? I'd try to re-run the same set of classes with
> the same master seed, without narrowing to that particular suite; may
> be a dependency issue.
>
>> Sorry, I am uninformed about randomizedrunner :-)
>
> Not a problem. I'll try to provide feedback; ping me directly if I
> miss an e-mail or a problem (it can happen given the amount of email
> junk I get).
>
> Dawid
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
---------------------------------------------------------------------

Hi,

This is not the case, ignore my previous mail!

The test method with the problem is using Solr's API, so I have no idea
which codecs or whatever it uses. It seems like a starvation/deadlock
problem. But all CPUs are tied 100% and my computer (also mouse cursor
stands still).

I opened https://issues.apache.org/jira/browse/SOLR-3461 and disabled the
test for now.

Uwe

> OK, found out:
>
> The test runs endless also on my local machine, what happened:
>
> - SerialMergeScheduler
> - SimpleText codec
>
> That says everything. I have no idea why it is faster on Linux, but that
> is a quite common config I get here also with other seeds.
>
> So we should annotate that test with the SuppressCodecs() annotation. Any
> ideas what codecs to suppress?
>
>>> How top get the seed? The test passed and I only see the ANT output
>>> like
>>> listed in the above URL. I only see the master seed, does this help to
>>> execute this one test isolated?
>>
>> The master seed is what matters. Everything else is a derivation of
>> that -- if you filter (narrow down to a particular method) that method
>> will still get an identical starting seed derived from the master.
>> That's also the reason the master seed is always printed by junit4 as
>> in here:
>>
>>     [mkdir] Created dir:
>> C:\Jenkins\workspace\Windows-Lucene-Solr-tests-only-trunk-java7\solr\build\solr-core\test
>>    [junit4] <JUnit4> says aloha! Master seed: 4877EC7806E56AF3
>>
>> Note that the execution here:
>> http://jenkins.sd-datasolutions.de/job/Windows-Lucene-Solr-tests-only-trunk-java7/11/consoleText
>>
>> had errors prior to the "long" test -- maybe this is what's causing
>> the slowdown somehow? I'd try to re-run the same set of classes with
>> the same master seed, without narrowing to that particular suite; may
>> be a dependency issue.
>>
>>> Sorry, I am uninformed about randomizedrunner :-)
>>
>> Not a problem. I'll try to provide feedback; ping me directly if I
>> miss an e-mail or a problem (it can happen given the amount of email
>> junk I get).
>>
>> Dawid
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [EMAIL PROTECTED]
>> For additional commands, e-mail: [EMAIL PROTECTED]
>>
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------