Ok sorry, false alarm. I figured this out by cleaning out my .m2
repository, re-building just hadoop-client-minicluster, and then iterating
all resulting jars in the .m2 using jar -tf with grep to find which jar was
providing these classes. Turns out it was
com.github.spotbugs:spotbugs-annotations:jar:3.1.9, which I could then see
in dependency:tree as coming from zookeeper 3.4.14. My build environment
was setting the zookeeper.version property to 3.4.14, and it seems like
this was a new dependency added between the hadoop-default version of
3.4.13 and 14.

The build works with 3.4.13, so now I can decide whether to use that
version or update the invariant check accordingly.

On Fri, Jun 11, 2021 at 1:23 PM Bryan Beaudreault <bbeaudrea...@hubspot.com>
wrote:

> Hey all,
>
> I'm trying to build hadoop from source. I've checked out the
> rel/release-3.2.2 tag from github.com/apache/hadoop. The maven build gets
> most of the way through, but fails on hadoop-client-check-test-invariants
> with the following error:
>
> [ERROR] Found artifact with unexpected contents:
> '/hadoop/rpm/scratch/BUILD/hadoop-3.2.2-src/hadoop-client-modules/hadoop-client-minicluster/target/hadoop-client-minicluster-3.2.2.jar'
>     Please check the following and either correct the build or update
>     the allowed list with reasoning.
>
>     edu/
>     edu/umd/
>     edu/umd/cs/
>     edu/umd/cs/findbugs/
>     edu/umd/cs/findbugs/annotations/
>     edu/umd/cs/findbugs/annotations/NonNull.class
>     ... snip (there's more classes in this package) ...
>
> I know I could modify the shell script to allow this, or I could try to
> modify the pom to exclude it. But it seems weird that I'm hitting this at
> all. I have not made any changes to the source yet.
>
> Has anyone else hit this or have any idea why it's happening? I can't find
> any reference to "edu/umd" or "edu.umd" in the source. I'm guessing it's
> coming from a transitive dependency, but I tried running mvn
> dependency:tree and did not see any mention of this version of findbugs.
>
> Thanks!
>

Reply via email to