Hi Till,
Yes, that is pretty much how they do it.
The trick is to get access to shared data between Spark realm and H2O realm.
Prev gen they use Tachyon but in the latest stint they use run H2O in
the Spark executor to use the heap shared memory to do the trick.
I am trying to hook us up with H2O
I think we are in the similar situation as to Spark when graduating,
we just have to move the PRs we wanted to new github mirror.
- Henry
On Thu, Jan 8, 2015 at 9:34 AM, Robert Metzger wrote:
> Hi,
>
> as per https://issues.apache.org/jira/browse/INFRA-8909 infra finished
> moving Flink to a TLP
Marton, could you close this VOTE thread by replying to the original
email and append [CANCEL] in the subject line.
- Henry
On Thu, Jan 8, 2015 at 9:35 AM, Márton Balassi wrote:
> Cherry-picked and tested: found no duplicate dependencies in lib, yarn
> uberjar build goes without the mentioned wa
Till Rohrmann created FLINK-1376:
Summary: SubSlots are not properly released in case that a
TaskManager fatally fails, leaving the system in a corrupted state
Key: FLINK-1376
URL: https://issues.apache.org/jira/b
Cherry-picked and tested: found no duplicate dependencies in lib, yarn
uberjar build goes without the mentioned warns.
Travis tests are passing, pushing soon.
On Thu, Jan 8, 2015 at 4:57 PM, Stephan Ewen wrote:
> Nice.
>
> @Marton: As soon as as you are done, I make a pass over the licenses...
>
Hi,
as per https://issues.apache.org/jira/browse/INFRA-8909 infra finished
moving Flink to a TLP project.
Please let me know if you find any remaining issues or report them directly
to infra.
The website and mailing lists are now running with the new name:
flink.apache.org.
The mailinglists were
Robert Metzger created FLINK-1375:
-
Summary: Remove incubator references & rename to new project urls
Key: FLINK-1375
URL: https://issues.apache.org/jira/browse/FLINK-1375
Project: Flink
Issu
+1 for indicating the person currently working on the issue, we can just
open a JIRA issue for each of these. And we can clearly indicate that other
features are not being currently worked on.
How about indicating rough time goals (quarters) for issues that are
currently being worked on (of course
