Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 02:22
    kimbyungeun opened #4098
  • Dec 15 19:47

    Aaronontheweb on dev

    TypeExtensions.TypeQualifiedNam… (compare)

  • Dec 15 19:47
    Aaronontheweb closed #4071
  • Dec 15 19:47
    Aaronontheweb closed #3767
  • Dec 15 19:47
    Aaronontheweb labeled #3767
  • Dec 15 19:47
    Aaronontheweb labeled #3767
  • Dec 15 19:47
    Aaronontheweb milestoned #3767
  • Dec 15 19:44
    Aaronontheweb labeled #4097
  • Dec 15 19:44
    Aaronontheweb milestoned #4097
  • Dec 15 13:23
    Aaronontheweb commented #4096
  • Dec 15 13:22
    Aaronontheweb commented #4093
  • Dec 15 13:16
    ismaelhamed commented #4093
  • Dec 15 13:04
    ismaelhamed edited #4097
  • Dec 15 13:04
    ismaelhamed opened #4097
  • Dec 15 12:50
    ismaelhamed commented #4096
  • Dec 15 12:48
    ismaelhamed commented #4096
  • Dec 15 12:05
    Aaronontheweb commented #4096
  • Dec 15 11:43
    ismaelhamed commented #4096
  • Dec 14 19:13
    hwanders commented #4096
  • Dec 14 13:05
    IgorFedchenko commented #4085
jweimann
@jweimann
yep :)
Aaron Stannard
@Aaronontheweb
:+1:
I'm going to OSS that whole sample soon
working on smoothing out some issues with CPU utilization on the worker nodes
trying to introduce throttling there to stop them from maxing out on massive domains
for the exact diminishing returns reasons I was talking about earlier :p
Coinio
@Coinio
@Aaronontheweb I was testing with MaxNumberOfBalls at around 100,000-500,000 just for kicks (comparing it against a single threaded non-akka version). I had a previous design where I was passing in a Ball class from a giant list to a BallUpdateActor pool which only needed a pool size of around 10. That worked okay, but required the Ball class to have mutable properties for the BallUpdateActor to work on which didn't seem to nice. So I thought having individual BallActors which managed their own state would be better. Need a rethink I guess...
@jweimann That's currently all they are doing. Just updating position from the velocity on receiving an update message.
Emil Ingerslev
@emilingerslev
@Aaronontheweb didn't see your response before now. Join seems good, but will gossip spread likewise, using no seed notes? Like if B joins C and A joins B, then all nodes know when A leaves, etc.?
Geert Van Laethem
@GeertVL
The LastSender in the TestKitBase is that also the replier when you do a actor.AskAndWait() ??
Suhas Chatekar
@schatekar
I was going though the docs and notices that there is lot of explanation of concepts which is good but code samples are missing
is this on purpose or a general lack of contribution in this area?
Bartosz Sypytkowski
@Horusiath
guys, your The Inevitable Rise of the Stateful Web Application post reached the latest highscalability.com
Roger Johansson
@rogeralsing
btw. we just reached 1100 stars on GitHub.. exactly the same time as Orleans :P
Aaron Stannard
@Aaronontheweb
guys, your The Inevitable Rise of the Stateful Web Application post reached the latest highscalability.com
:sparkles:
@schatekar I'd chalk it up to a combination of lack of attention / not knowing what should be included as an example there
the docs definitely need some more love
@emilingerslev yep, gossip will still spread - as long as you can connect to a node that's already in the cluster, you're good to go
Aaron Stannard
@Aaronontheweb
seed nodes aren't special - they're just the first nodes that new nodes contact when they attempt to join
@Coinio ah, so under that scenario, since you're using a broadcast router, all 100,000-500,000 balls are going to attempt to update at once
still though, that's pretty cool :p
@rogeralsing hehe - nice.
Emil Ingerslev
@emilingerslev
@Aaronontheweb I hoped that was true. I makes the setup very reliable, which is good. Thanks :)
jweimann
@jweimann
@Aaronontheweb @Coinio I don't think the broadcast router makes sense for those balls, unless they're all supposed to change velocity at the same exact second to the same value, but I doubt that's the expected case
Bartosz Sypytkowski
@Horusiath
with akka-jvm 2.4, there are a lot of changes in cluster sharding:
  1. They moved persistence backed for sharding coordinator from Akka.Persistence to CRDTs
  2. Protobufed messages
  3. Graceful shutdown
Aaron Stannard
@Aaronontheweb
quite the moving target
but I get it - that stuff's not trivial to implement
Birnsen
@Birnsen

Hey everyone. Fantastic project! Following a while now... Watched your cluster contrib meeting on YT and there you mentioned the need of a MultiNode Visulaizer.
So I did a little something :). It is not pretty and it is written in F# (currently learning so code is probably a mess), so do not expect too much.
But maybe I can do something out of it (spit out static HTML) if needed.

https://github.com/Birnsen/MultiNodeVisualizer/blob/master/mnv.png

The numbers are the amount of events at that given moment. If an test fails is also not implemented yet but should be easy to do...
Had only one file to test it so others will probably fail.

derrickcrowne
@derrickcrowne
Are there any docs or examples on how to use UdpConnected in Akka.IO? There are extension methods on ActorSystem for Udp() and Tcp() but I couldn't find one for UdpConnected().
derrickcrowne
@derrickcrowne
I tried sending the UdpConnected.Connect message to ActorSystem.Udp() but it said it only responds to Bind and SimpleSender
Joshua Benjamin
@annymsMthd
@Aaronontheweb @derrickcrowne is starting to use Akka.IO to build a vehicle simulator
Suhas Chatekar
@schatekar
@Birnsen is that something you are outputting on the console?
Birnsen
@Birnsen
@schatekar yes
but the json file as input not real time
Suhas Chatekar
@schatekar
How difficult would it be to make your program spit out an html that would render the same thing?
Birnsen
@Birnsen
not that much... all the data should be given. only have to do a the skeleton to put the data in.
i am not that much of a js guy though
Patrick Mohr
@kc7zzv
Does Akka have an equivalent to RxEx's .Buffer()? Buffer works by queuing up messages, but instead of sending them through as they arrive, it will queue up to n messages, or queue up messages for n seconds. When either the time or the capacity limit is reached, it sends through all the messages at once. For example, I might put "buffer 5 seconds" in front of writing changes to the cache to disk. That way I can merge the changes into a bigger write. (Sacrificing up to 5 seconds of data if the program crashes.)
I could probably build this using an Actor, and a delayed message send using the scheduler, but I'm guessing something like this already exists, if I just knew the name.
Chris Martin
@trbngr
@Aaronontheweb playing with Akka.Persistence.Cassandra but can't get tests to run. NullRef serializing a Persistent with an int payload
everything is up to date
Aaron Stannard
@Aaronontheweb
@trbngr yeah, we have a bug logged here akkadotnet/akka.net#1235
not 100% sure what's going on there - haven't investigated yet, but it looks like the Cassandra plugin was depending on something for serialization that the other plugins werent
Chris Martin
@trbngr
well balls
it's weird man. I can pull the serialization code into the Cassandra project and it serializes just fine
Chris Martin
@trbngr
PR sent
Hyungho Ko
@hhko
how to disconnect remote ActorSystem successfully. normally. remote ActorSystem print out error and warning messages stdout. I want to disconnect it without error and warning messages.
Hyungho Ko
@hhko
how to handle disconnection event from client AcorSystem in Server AcorSystem?