Just to be clear, the stream is of String elements. The first part of the
pipeline (up to the first .apply) receives those strings, and returns
objects of another class ("A" let's say).
On Thu, Dec 22, 2016 at 6:04 PM, Matt wrote:
> Hello,
>
> I have a window processing 10 objects at a time, and
Hello,
I have a window processing 10 objects at a time, and creating 1 as a
result. The problem is in order to create that object I need the object
from the previous window.
I'm doing this:
stream
.keyBy(...some key...)
.countWindow(10, 1)
.apply(...creates an element A...)
.keyBy(...sam
Aljoscha,
For the second, possible solution is there also a requirement that the data
sinks handle out-of-order writes? If the new job outpaces the old job which
is then terminated, the final write from the old job could have overwritten
"newer" writes from the new job.
Greg
On Tue, Dec 20, 2016
Looking at the Kafka 0.8 connector API, my deserializer definitely gets the
message but none of the header information, and in particular, message
metadata.
Is there a straightforward way to do this (other than upgrading our Kafka
cluster ;)). The connector code itself is a bit involved.
Ron
Hi Stephan -
I agree that the savepoint-shutdown-restart model is nominally the same as the
rolling restart with one notable exception - a lack of atomicity. There is a
gap between invoking the savepoint command and the shutdown command. My problem
isn’t fortunate enough to have idempotent oper
Hi,
I'd like to know which is more efficient: splitting a stream 3 ways via `split`
or via `filter`?
--- FILTER --
val greater = stream.filter(_.n > 0)
val less = stream.filter(_.n < 0)
val equal = stream.filter(_.n == 0)
-
- VS -
--- SPLIT ---
val split = stream.split(
Hi,
I noticed the following Problem with a POJO I use to encapsulate Values.
java.lang.NullPointerException
at
org.joda.time.tz.CachedDateTimeZone.getInfo(CachedDateTimeZone.java:143)
at
org.joda.time.tz.CachedDateTimeZone.getOffset(CachedDateTimeZone.java:103)
at
org.j
Here is the code of a Double wrapper with null support [1].
[1] https://gist.github.com/a8e8aa377957d3d51eadf36fe5c92a9e
On Tue, Dec 20, 2016 at 4:26 PM, Anirudh Mallem
wrote:
> If you are using Avro generated classes then you cannot have your values
> null.
> https://cwiki.apache.org/confluenc
That approach should work as well.
The upcoming Flink 1.2.0 release will feature a function for asynchronous
operations, i.e., you can have multiple concurrent Redis requests, without
losing the fault tolerance guarantees.
Another alternative is to store the map in key-partitioned operator state
o
