I am using akka-stream-kafka 0.18. I have a flow that reads from one kafka
topic, does some processing and then writes to a different kafka topic. The
flow has been shutting down intermittently when kafka brokers fail.
Sometimes the brokers will fail repeatedly over a long period and the flow
Responding to Martynas Mickevičius:
> message send will be indeed in a race with system termination. A better
> approach would be to use an ? (ask) and have the actor under `actorUrl` to
> send a response when everything is sent and done.
Ah, nice, didn't think of that for some reason - that
Please take a look at
https://github.com/paypal/squbs/blob/master/squbs-unicomplex/src/main/scala/org/squbs/unicomplex/OrderingStage.scala
for
ordering. I will be making changes in the next few weeks to put a limit on
the internal buffer and make it more lenient with a flag (
Yes, 9 nodes in 1 cluster
On Friday, January 12, 2018 at 12:47:21 PM UTC+5:30, j...@spineor.com wrote:
>
> Hi
>
> I have just started using Akka and creating a demo application using Akka
> persistence. This application has only one actor persisting event in
> cassandra. I am running this
Hi,
not sure if you have solved this already, but one idea comes to my mind,
where your partitioner would inject a marker which would tell which part of
the stream you need to consume next when merging. This means that at the
partitioner side you would have to buffer at least two elements (hold
A PersistentActor with a given persistenceId can only be active at one
place (single writer principle). Cluster Sharding will help you with that.
/Patrik
fre 12 jan. 2018 kl. 09:08 skrev Martynas Mickevičius <
martynas.mickevic...@lightbend.com>:
> Hi,
>
> are your Cassandra nodes in one
Hi,
are your Cassandra nodes in one cluster?
On Fri, Jan 12, 2018 at 2:17 PM wrote:
> Hi
>
> I have just started using Akka and creating a demo application using Akka
> persistence. This application has only one actor persisting event in
> cassandra. I am running this
Hi,
if you complete a request with a `Source` you can attach cleanup actions
using watchTermination.
On Thu, Jan 11, 2018 at 11:55 PM Daniel Armak wrote:
> Is there a way to be notified when a response has been sent to the client?
>
> I’d like to cleanup some resources used
Hi,
message send will be indeed in a race with system termination. A better
approach would be to use an ? (ask) and have the actor under `actorUrl` to
send a response when everything is sent and done.
On Thu, Jan 11, 2018 at 7:58 PM Patrick Roemer
wrote:
> Hi,
>
>