Skip to content
This repository has been archived by the owner on Nov 9, 2019. It is now read-only.

CalculateCoverageComponentsIntegrationTest.testSimpleMatrix fails intermittently #242

Closed
lbergelson opened this issue Dec 10, 2015 · 7 comments
Assignees

Comments

@lbergelson
Copy link
Member

org.broadinstitute.hellbender.tools.exome.CalculateCoverageComponentsIntegrationTest.testSimpleMatrix FAILED
    java.lang.AssertionError: expected [4] but found [3]
        at org.testng.Assert.fail(Assert.java:94)
        at org.testng.Assert.failNotEquals(Assert.java:496)
        at org.testng.Assert.assertEquals(Assert.java:125)
        at org.testng.Assert.assertEquals(Assert.java:372)
        at org.testng.Assert.assertEquals(Assert.java:382)
        at org.broadinstitute.hellbender.utils.pca.PCAUnitTest.assertEquals(PCAUnitTest.java:201)
        at org.broadinstitute.hellbender.tools.exome.CalculateCoverageComponentsIntegrationTest.testSimpleMatrix(CalculateCoverageComponentsIntegrationTest.java:54)

This was seen during rebuilds of #192

@lbergelson
Copy link
Member Author

@vruano I made a ticket for this just in case others run into it. Feel free to close if you think it's not relevant.

@vruano
Copy link
Contributor

vruano commented Dec 10, 2015

Interestingly I cannot reproduce the error and master is currently passing tests.

@vruano vruano self-assigned this Dec 10, 2015
@vruano
Copy link
Contributor

vruano commented Dec 10, 2015

I tried to disable Spark on that tool which is on by default but either-way it passes the tests.

@lbergelson
Copy link
Member Author

Must be some race condition in the tests then? Or maybe something platform specific to travis?

@vruano
Copy link
Contributor

vruano commented Dec 10, 2015

So what should I do?! I can disable spark in the test to avoid future recurrences yet that is sweeping it under the carpet.

@vruano
Copy link
Contributor

vruano commented Dec 10, 2015

Btw this seems to affect @davidbenjamin issue as well as it passes in master.

vruano added a commit that referenced this issue Dec 11, 2015
…using Spark and another

using Apache common for SVD.

This is in response to a difficult to reproduce error when testing in Travis. If the Spark
version is the one that has the problems it makes it more likely to be a race-condition problem.

The problem has been reported in issue #242

Fixes bug in CalculateCoverageComponents  reported in #Blah.

It seems that this is due to relay on the Spark based SVD implementation when running in Travis.
Disabling spark solves the issue.
vruano added a commit that referenced this issue Dec 14, 2015
…using Spark and another

using Apache common for SVD.

This is in response to a difficult to reproduce error when testing in Travis. If the Spark
version is the one that has the problems it makes it more likely to be a race-condition problem.

The problem has been reported in issue #242

After a few Travis re-run the problem only occurs when using Spark SVD.

For now I disable the possibility of using Spark for the SVD/PCA.
vruano added a commit that referenced this issue Dec 14, 2015
…using Spark and another

using Apache common for SVD.

This is in response to a difficult to reproduce error when testing in Travis. If the Spark
version is the one that has the problems it makes it more likely to be a race-condition problem.

The problem has been reported in issue #242

After a few Travis re-run the problem only occurs when using Spark SVD.

For now I disable the possibility of using Spark for the SVD/PCA.
vruano added a commit that referenced this issue Dec 15, 2015
…using Spark and another

using Apache common for SVD.

This is in response to a difficult to reproduce error when testing in Travis. If the Spark
version is the one that has the problems it makes it more likely to be a race-condition problem.

The problem has been reported in issue #242

After a few Travis re-run the problem only occurs when using Spark SVD.

For now I disable the possibility of using Spark for the SVD/PCA.
@vruano
Copy link
Contributor

vruano commented Feb 24, 2016

Will close and it should have been fixed. Please open if it reappears.

@vruano vruano closed this as completed Feb 24, 2016
lbergelson pushed a commit that referenced this issue May 31, 2017
…using Spark and another

using Apache common for SVD.

This is in response to a difficult to reproduce error when testing in Travis. If the Spark
version is the one that has the problems it makes it more likely to be a race-condition problem.

The problem has been reported in issue #242

After a few Travis re-run the problem only occurs when using Spark SVD.

For now I disable the possibility of using Spark for the SVD/PCA.
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants