If you don't want to write to an actual file, the example with the Check
transform should allow you to use Check(...) as you would a sink. (I
realize this should have been
run_my_pipeline(
beam.Create([...]),
"Check1" >> Check(equal_to([...])),
"Check2" >>
Hello Robert
could you point me to a test sample where a 'mock' sink is used?
do you guys have a testing package , which provide an in memory sink where
for example i can dump the result of
my pipeline (as opposed to writing to a file) ?
Additionally, what is the best way to test writing to
Hi Alexey,
I did explore that route. The problem there is to identify what timestamp to
use. While processing records, you can capture the timestamp. This cannot be
processing time, but event time on the Kinesis record. As far I see, event time
on Kinesis record is generated from
Hi Mani,
Knowing when you last run was stopped (since you use batch mode), could you
leverage “withInitialPositionInStream()” or “withInitialTimestampInStream()”
for KinesisIO in this case?
Alexey
> On 21 Jul 2020, at 13:40, Sunny, Mani Kolbe wrote:
>
> Hi Max,
>
> Thank you for your
Hi Max,
Thank you for your reply. Our use case is to run a batch job against a Kinesis
source. Our downstream systems are still on batch mode. So this application
will read from the Kinesis source periodically and generate batched outputs.
Without ability to resume from a checkpoint, it will
Hi Mani,
BoundedReadFromUnboundedSource was originally intended to be used in
batch pipelines. In batch, runners typically do not perform
checkpointing. In case of failures, they re-run the entire pipeline.
Keep in mind that, even with checkpointing, reading for a finite time in
the