+1

- Verified all checksums and signatures.
- Checked LICENSE and NOTICE.
- Compared the files in the src tarball with the files in git.
- Built from src tarball successfully.

- Reproduced the files in the bin tarball:
-* It turns out that the jars are unsigned.
-* Except for ratis-proto-3.2.2.jar, all the other jars in the rc can
be reproduced exactly.
-* The file ratis-proto-3.2.2.jar in rc somehow has an empty
"annotations/" directory but my generated jar doesn't.
-* After extracting the jar, all the files inside the jar are exactly
the same as the files generated locally.

- Passed all unit tests.
- Able to run examples and shell with the bin tarball.


Thanks for rolling out the RC!

Tsz-Wo


On Mon, Mar 30, 2026 at 10:45 AM Tsz Wo Sze <[email protected]> wrote:

> Hi Xinyu,
>
> Thanks for the info!  I got the same MD5 now:
>
> % md5
> ratis-common/target/classes/org/apache/ratis/conf/RaftProperties.class
> MD5
> (ratis-common/target/classes/org/apache/ratis/conf/RaftProperties.class) =
> faf62bea5c84206da70dfa626b03b004
>
> Will continue verifying the release.
>
> Tsz-Wo
>
> On Sat, Mar 28, 2026 at 9:58 PM Xinyu Tan <[email protected]> wrote:
>
>> Hi, Tsz-wo
>>
>> Sorry for the late reply.
>>
>> I'm using jdk 1.8.0_472 with Zulu 8.90.0.19-CA-macos-aarch64.
>>
>> Best
>> ----------------
>> Xinyu Tan
>>
>> On 2026/03/27 17:22:26 Tsz Wo Sze wrote:
>> > Hi Xinyu,
>> >
>> > What is the jdk version you used to build the RCs?
>> >
>> > Extracted the jars and then compared the class file.  However, the class
>> > files are different.  I probably need to use the exact jdk to reproduce
>> the
>> > binaries.
>> >
>> > % from the bin tarball
>> > MD5
>> >
>> (/Users/szetszwo/ratis/release/3.2.2/rc3/apache-ratis-3.2.2-bin/tmp/jars/ratis-common-3.2.2/org/apache/ratis/conf/RaftProperties.class)
>> > = faf62bea5c84206da70dfa626b03b004
>> >
>> > % generated locally using make_rc.sh
>> > MD5
>> (jars/ratis-common-3.2.2/org/apache/ratis/conf/RaftProperties.class) =
>> > 7a35e5b500c4cf6283ce705f4bbf31a3
>> >
>> > Tsz-Wo
>> >
>> >
>> > On Thu, Mar 26, 2026 at 8:16 PM Tsz Wo Sze <[email protected]> wrote:
>> >
>> > > I am verifying the rc3.
>> > >
>> > > Question: How to verify if the binary is reproducible?
>> > > - I rebuilt the tarball.  However, the resulting binary was different
>> > > since my generated jars were signed by my signatures.
>> > > - Also tried the diffoscope tool recommended by [1] but it showed a
>> lot of
>> > > difference (due to the signatures?).
>> > >
>> > > Tsz-Wo
>> > > [1] https://reproducible-builds.org/tools/
>> > >
>> > > On Sun, Mar 22, 2026 at 2:12 AM Attila Doroszlai <
>> [email protected]>
>> > > wrote:
>> > >
>> > >> > I’m calling a vote for Apache Ratis Release 3.2.2 rc3.
>> > >> > https://github.com/apache/ratis/tree/ratis-3.2.2-rc3
>> > >> > https://dist.apache.org/repos/dist/dev/ratis/3.2.2/rc3
>> > >> >
>> https://repository.apache.org/content/repositories/orgapacheratis-1178
>> > >>
>> > >> +1
>> > >>
>> > >> - Verified checksums, signatures, git hash
>> > >> - Compared source tarball contents to git repo
>> > >> - Built from source
>> > >> - Ozone CI mostly passed (except a single test) with staged Maven
>> > >> artifacts
>> > >>
>> > >> Thanks Xinyu for the RC (and fixing issues with previous RCs).
>> > >>
>> > >> -Attila
>> > >>
>> > >
>> >
>>
>

Reply via email to