Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 11:24
    Kenji-Tanaka commented #3887
  • 10:45
    nagytech edited #4086
  • 10:45
    nagytech synchronize #4086
  • 10:43
    nagytech opened #4086
  • 09:00
    Horusiath commented #4077
  • 06:31
    Aaronontheweb commented #4085
  • 06:28
    Aaronontheweb commented #4085
  • 06:24
    Aaronontheweb commented #4085
  • Dec 07 11:49
    IgorFedchenko commented #4085
  • Dec 07 10:31
    IgorFedchenko commented #4085
  • Dec 07 08:36
    IgorFedchenko commented #4085
  • Dec 06 20:57
    IgorFedchenko commented #4085
  • Dec 06 20:32
    IgorFedchenko commented #4085
  • Dec 06 20:01
    IgorFedchenko commented #4085
  • Dec 06 19:55
    IgorFedchenko commented #4085
  • Dec 06 16:22
    Aaronontheweb labeled #3997
  • Dec 06 16:22
    Aaronontheweb closed #3997
  • Dec 06 16:20
    IgorFedchenko commented #3997
  • Dec 06 16:08
    IgorFedchenko commented #4085
  • Dec 06 15:50
    Aaronontheweb assigned #4085
Bartosz Sypytkowski
@Horusiath
@rogeralsing nope, just for some specs (I've noticed that one of my tests using TestLatches was hanging when message serialization was turned on)
but eventually I just moved it to separate spec class
Roger Johansson
@rogeralsing
I've seen a similar problem before, back then, it was that some static resources, like NoRouter or DefaultDeploy or whatver, was overwritten by the serializer .. e.g. some primitives took a static resource as an input argument, then the serializer overwrote the properties of the primitive, which was backed by the static resource, and tada , the static resource was corrupted
could that be a similar problem here?
e.g. public Props() : this(Some.Static.Resource) {} then when the deserializer tried to deserialize an object with such ctor, the injected resource would be corrupted
Bartosz Sypytkowski
@Horusiath
I know that test latch uses System.Threading.CountdownEvent so I supose it may be not safe to try to serialize/deserialize it
Roger Johansson
@rogeralsing
were you passing a testlatch object as a message?
Bartosz Sypytkowski
@Horusiath
I have to for test verification ;)
Roger Johansson
@rogeralsing
and if so, shouldnt that be marked with the INoSerializationNeeded or whatever the name
NoSerializationVerificationNeeded
that will ofc not help if the testlatch is part of a bigger message, but if it is the root, that would make it bypass serialization in inproc systems
Bartosz Sypytkowski
@Horusiath
it worked, I didn't know about that interface
Roger Johansson
@rogeralsing
its in the scala version too
neat :) so for persistence, we just need to start making providers? and follow typesafes lead on how to take it out of beta (?)
Bartosz Sypytkowski
@Horusiath
not all of exisiting tests are passing yet (6 out of 82 total are marked to be skipped, I need to fix them probably)
beside that, I'm currently during writing some tutorial about how to write custom persistence plugins
Roger Johansson
@rogeralsing
So what probiders do you think are the most important ones? SqlServer, different Azure storages, flat file?
Bartosz Sypytkowski
@Horusiath
also I'm in midway of creating Akka.Persistence.SqlServer plugin
Roger Johansson
@rogeralsing
:+1:
Bartosz Sypytkowski
@Horusiath
those two definitely yes, snapshot store for local file system already exists (I've rewritten it <not pushed yet>). I don't know if some other providers should be included to akka.net core repository
but I think, that @Aaronontheweb or @jcwrequests may want to try to implement Cassandra plugins for Akka.Persistence
I also hope, that when it will be ready to drop -Pre flag, RavenDB and EventStore creators will want to integrate their services too
jcwrequests
@jcwrequests
@horusiath
jcwrequests
@jcwrequests
@horusiath is your SQL code on your Akka fork? I think it would be a good idea to write them for Cassandra, mongo, postgres and maybe foundation db. I think that the focus should be on the high write throughput DBs. @arronontheweb might have some during opinions about ravendb 😊.
Bartosz Sypytkowski
@Horusiath
@jcwrequests if you want to start, my sqlserver fork is not what you want (at least for now). Right now I'm hardening Akka.Persistence.Tests (they are randomly failing right now, mostly due to timeouts) and making an Akka.Persistence.TestKit to work easier with. If everything goes right, they could be released soon and then I'll prepare tutorial about how to start persistence plugin development - right now I already have sample code, but it's not commented.
Natan Vivo
@nvivo
@all, general question: in which cases Akka doesn't own the threads it uses for actors? And are these scenarios recommended for production or just for testing?
Roger Johansson
@rogeralsing
By default we execute on the threadpool, so we don't own those threads.. we can also execute on the GUI thread with the synchronization context dispatcher if you want an actor that updates GUI elements. we dont own that thread either... then we have cases like the new threadpool, that can be used to give some actors a higher priority than the others, those threads are owned by whatever threadpool we use under the hood, but yes you could consider those owned threads I guess.. so we can execute ontop of whatever the current dispatcher decides to run the work on
but in short, akka never does anything to the threads, it just executes the mailbox run on whatever scheduling mechanism is there
Natan Vivo
@nvivo
@rogeralsing got it. thanks!
jcwrequests
@jcwrequests
@Horusiath Got it.
Stefan Sedich
@stefansedich
When using PipeTo and an exception is raised in the Task is it possible to get that to propogate up to an unhandled actor exception so that it goes to supervision?
Stefan Sedich
@stefansedich
@Aaronontheweb answered the above for me, it was too late in the day and I was derping a bit.
I do have a question for the group though on how you would handle supervision in this scenario: https://gist.github.com/stefansedich/0c5e1d2ac431eccd9fed
In this scenario say DoStuff failed, I would want to restart the child, but to do that I would need to execute the Start message again to reset the internal state back to a starting point before kicking off again, what are the best ways to achieve this kind of thing?
My current solution is to pass a message to the parent, but I end up implementing my own retry and not taking advantage of the Supervisor strategy.
Roger Johansson
@rogeralsing
And I assume that Start carries information that is owned by someone else, e.g. the Parent here? so the parent have to be the one passing it?
Stefan Sedich
@stefansedich
yeah
sorry if the example is a bit shitty :)
the child actually notifies the parent of the change and parent keeps track of state, so it it has to restart parent would need to pass down the current value.
It is a position in a stream basically.
Roger Johansson
@rogeralsing
The actor logging solves this by passing back a message to the parent notifying that it is ready to work.. so in the ChildActor ctor, you could add a "Parent.Tell(new IAmReadyToStart())" kind of message, or is that what you are already doing?
when parent receives the IAmREadyToWork, it can pass the Start message back to the child
Stefan Sedich
@stefansedich
yeah that happens, the issue is how does supervision handle when DoWork fails and the parent has to pass a new Start after the child restarts.
Roger Johansson
@rogeralsing
the child will send a new IAmReadyToWork to the parent.. if you place that in the ctor
as the child actor will be recreated if you use Restart
Stefan Sedich
@stefansedich
so basically bypass supervision then?
Roger Johansson
@rogeralsing
no
Stefan Sedich
@stefansedich
oh right yep sorry
Roger Johansson
@rogeralsing
the ctor is always called on the restarting actor. so you can place parent notifications in there
Stefan Sedich
@stefansedich
yeah so the trick is let the child ask for work