How much more work would be involved in generating a separate set of
Hadoop 3 packages for those who would like to try out HDFS erasure
coding (among other things)?
Thanks,
Dan
On 08/07/2018 10:04 PM, Jun HE wrote:
Thanks for your inputs.
For failing packages I agree with Olaf and Evans: remove them if no one
would take over the fix task.
Oozie and pig have JIRAs(BIGTOP-2986/BIGTOP-2975). For crunch there isn't.
I'll create one later and do some initial investigation to see if I can fix
it or not.
For Hadoop upgrade, I'm fine to upgrade it to 2.8.4. Actually we've done
this in Linaro's ERP-18.06 release. (Yes, we're using Bigtop to build
BigData packages :)) So there is no problem for it.
Regards,
Jun
2018-08-08 10:31 GMT+08:00 Evans Ye <[email protected]>:
Yes. We've features/bugfixes stacked up over time and definitely need a
release.
I'm +1 to what Olaf said. Unless someone is willing to take over the task
and fix it.
From my perspective, Oozie and Pig's market are declining. So I suggest
let's spend our effort on those major components, and the functionality
that Bigtop provides such as deployment and testing.
Naresh Bhat <[email protected]> 於 2018年8月8日 週三 上午8:58寫道：
+1
On 7 August 2018 at 21:41, Jun HE <[email protected]> wrote:
Hi, folks,
As you may read from previous thread, I'm proposing to do a new
release,
1.3.0, before ApacheCon in Sept.
I'm a +1 for this because:
* quite a few fixes/improvements for building and deployment
* important components upgrade
* x86 and non-x86 platforms are aligned in master
If there are enough votes for a release I'd like to volunteer to be the
RM
for 1.3.0.
Evans has initiated a well-defined release BOM (
https://issues.apache.org/jira/browse/BIGTOP-2945), and I did some
minor
modifications mainly about components versions in 1.3.0.
And issues might need fix: https://s.apache.org/Ja7e
Your comments/thoughts are mostly welcomed!
Regards,
Jun
