[
https://issues.apache.org/jira/browse/CASSANDRA-17884?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17610019#comment-17610019
]
Ekaterina Dimitrova edited comment on CASSANDRA-17884 at 9/27/22 12:53 PM:
---------------------------------------------------------------------------
-I actually tried an unreleased in maven central jamm jar with the mid
resources and this time I didn't see the issue. I will try one more run to
confirm that the issue was solved and I am not missing anything.-
This is the
[branch|https://github.com/apache/cassandra/compare/trunk...ekaterinadimitrova2:cassandra:17884-4.0-hack]
where I added a jar including [this
commit|https://github.com/jbellis/jamm/commit/33d78c1035211e5f9e69ad9247f3a0f6ef75c923]
(this is not the latest work in the repo though) which was supposed to solve
that issues with JDK11 according to the author commit message in the jamm repo.
It seems it didn't and we are missing something else to filter. -
[https://app.circleci.com/pipelines/github/ekaterinadimitrova2/cassandra/1955/workflows/20b3aa2c-fefd-42df-8ad0-b787add2fbb1/jobs/15494/steps]
I added annoying print in the jamm code and verified it is printed in the
CircleCI logs to ensure nothing went wrong and we use the new jar I plugged in.
I also think we need to get back to the repo
[TreePrint|https://github.com/ekaterinadimitrova2/jamm/blob/33d78c1035211e5f9e69ad9247f3a0f6ef75c923/src/org/github/jamm/TreePrinter.java]
class as I think it is really useful for debugging. Others think in the same
way. If we feel it lacks some testing, better to add that than to remove it
completely.
was (Author: e.dimitrova):
I actually tried an unreleased in maven central jamm jar with the mid resources
and this time I didn't see the issue. I will try one more run to confirm that
the issue was solved and I am not missing anything.
This is the
[branch|https://github.com/apache/cassandra/compare/trunk...ekaterinadimitrova2:cassandra:17884-4.0-hack]
where I added a jar including [this
commit|https://github.com/jbellis/jamm/commit/33d78c1035211e5f9e69ad9247f3a0f6ef75c923]
(this is not the latest work in the repo though) which was supposed to solve
that issues with JDK11 according to the author commit message in the jamm repo.
It seems it didn't and we are missing something else to filter. -
https://app.circleci.com/pipelines/github/ekaterinadimitrova2/cassandra/1955/workflows/20b3aa2c-fefd-42df-8ad0-b787add2fbb1/jobs/15494/steps
I added annoying print in the jamm code and verified it is printed in the
CircleCI logs to ensure nothing went wrong and we use the new jar I plugged in.
I also think we need to get back to the repo
[TreePrint|https://github.com/ekaterinadimitrova2/jamm/blob/33d78c1035211e5f9e69ad9247f3a0f6ef75c923/src/org/github/jamm/TreePrinter.java]
class as I think it is really useful for debugging. Others think in the same
way. If we feel it lacks some testing, better to add that than to remove it
completely.
> Test failure: o.a.c.repair.RepairJobTest.testNoTreesRetainedAfterDifference
> ---------------------------------------------------------------------------
>
> Key: CASSANDRA-17884
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17884
> Project: Cassandra
> Issue Type: Bug
> Components: Test/unit
> Reporter: Andres de la Peña
> Priority: Normal
> Fix For: 4.0.x, 4.1.x, 4.x
>
>
> The unit test
> {{org.apache.cassandra.repair.RepairJobTest#testNoTreesRetainedAfterDifference}}
> seems to be slightly flaky at least in 4.0. We haven't seen it failing on
> Butler, but after [a failure on a patch
> branch|https://app.circleci.com/pipelines/github/jacek-lewandowski/cassandra/281/workflows/e6094c00-b611-4772-9772-205f4c76ecba/jobs/2126/tests],
> this repeated run shows a 0.28% flakiness:
> *
> https://app.circleci.com/pipelines/github/adelapena/cassandra/2076/workflows/8adbfe99-afb5-43af-84ad-43df2a2a86e2/jobs/20816/tests
> *
> https://app.circleci.com/pipelines/github/adelapena/cassandra/2076/workflows/e550d8c2-7c35-4326-bd95-6909978d344b/jobs/20817/tests
> {code}
> java.lang.reflect.InaccessibleObjectException: Unable to make field private
> jdk.internal.platform.cgroupv1.SubSystem$MemorySubSystem
> jdk.internal.platform.cgroupv1.Metrics.memory accessible: module java.base
> does not "opens jdk.internal.platform.cgroupv1" to unnamed module @157853da
> at
> java.base/java.lang.reflect.AccessibleObject.checkCanSetAccessible(AccessibleObject.java:340)
> at
> java.base/java.lang.reflect.AccessibleObject.checkCanSetAccessible(AccessibleObject.java:280)
> at
> java.base/java.lang.reflect.Field.checkCanSetAccessible(Field.java:176)
> at java.base/java.lang.reflect.Field.setAccessible(Field.java:170)
> at org.github.jamm.MemoryMeter.addFieldChildren(MemoryMeter.java:330)
> at org.github.jamm.MemoryMeter.measureDeep(MemoryMeter.java:269)
> at
> org.apache.cassandra.utils.ObjectSizes.measureDeep(ObjectSizes.java:216)
> at
> org.apache.cassandra.repair.RepairJobTest.testNoTreesRetainedAfterDifference(RepairJobTest.java:271)
> at
> java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at
> java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
> at
> java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> {code}
> Flakiness is really low but the failure is clearly reproducible in j11.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
