This vote will remain open for 96 hours (4 days, due to the weekend, and because I know some people want to finish some long tests), until Tue, April 15, 23:00 UTC 2014. (That's 7pm EDT.)
[ ] +1 - I have verified and accept... [ ] +0 - I have reservations, but not strong enough to vote against... [ ] -1 - Because..., I do not accept... ... these artifacts as the 1.6.0 release of Apache Accumulo.
This file has an incomplete license: core/src/test/resources/shelltest.txt but I don't think that's a blocker due to the trivial contents of the file.
The src tarball matches the git branch.
There are no javadocs in the bin tarball. There is no pre-built native map in the bin tarball (but I built one easily using the provided script). Is that expected? On Fri, Apr 11, 2014 at 3:41 PM, Christopher <[EMAIL PROTECTED]> wrote:
On Mon, Apr 14, 2014 at 4:02 PM, Billie Rinaldi <[EMAIL PROTECTED]> wrote:
Right, I stripped them out as part of https://issues.apache.org/jira/browse/ACCUMULO-1487 This is expected. This is also expected. It's easier to provide a platform-independent artifact that can be easily built, than guess at the target environment and produce something that will likely have to be recompiled anyway.
-1 due to the large performance regression (~40% degradation) identified by Jonathan Park in ACCUMULO-2668
Additionally, it would be nice to get ACCUMULO-2665 to ensure that we can unit test against Apache Hadoop 2.4.0 for 1.6.0, but I am not -1 on that issue alone (just -0 if it comes to that).
Aside from those two, the rest of the release looks pretty good to me:
* Used 86a9a8a1 to run unit and integration tests against Apache Hadoop 2.2.0, 2.3.0, and 2.4.0 (with ACCUMULO-2665 caveat for 2.4.0) * Verified sigs and hashes for tarballs (didn't write a script to verify all sigs from the staging repo -- I trust nexus here) * Ran short CI+verification on Apache Hadoop 2.3.0 * Ran short CI on Apache Hadoop 2.4.0 (ran into an issue with launching the verification job -- maybe it's just me doing something dumb?) * Verified basic compatibility with Apache Pig 0.13.0-SNAPSHOT (caveat PIG-3885). Read/Write data, join two tables. * Downloaded source tarball, built jars and nativemaps, ran unit and integration tests, and tested basic read/write test on Apache Hadoop 2.3.0 * Downloaded binary tarball, built nativemaps, tested basic read/write on Apache Hadoop 2.3.0
One final (repeated) comment, the binary tarball lack of a prebuilt native map. I believe this merits mention in the release notes before those are finalized. I will update this.
I don't believe we'll have a 24hr RandomWalk run of the LongClean module with agitation done by the time this vote closes. Since that's in our release guide, I think it should be done prior to a release.
If there's a run of this test going that we haven't heard about yet, I'd be happy to reconsider my vote. On Tue, Apr 15, 2014 at 12:14 AM, Josh Elser <[EMAIL PROTECTED]> wrote: