Hi folks!
I'm trying to implement an update of a broadcast var in Spark Streaming.
The idea is that whenever some configuration value has changed (this is
periodically checked by the driver) the existing broadcast variable is
unpersisted and then (re-)broadcasted.
In a local test setup (usi
I don't think that can work. Your BroadcastUpdater is copied to the task,
with a reference to an initial broadcast. When that is later updated on the
driver, this does not affect the broadcast inside the copy in the tasks.
On Wed, Sep 28, 2022 at 10:11 AM Dipl.-Inf. Rico Bergmann <
i...@ricobergma
hi,all :
when I try to merge a iceberg table by spark , I can see faild job on
spark ui , but the spark application final state is SUCCEEDED.
I submit an issue : https://github.com/apache/iceberg/issues/5876
I wonder to know is this a real error ? thanks .