These are chat archives for akkadotnet/akka.net

16th
Sep 2016
Damian Reeves
@DamianReeves
Sep 16 2016 00:51
Can my ActorSystem be run in Azure in a Web job?
qwoz
@qwoz
Sep 16 2016 01:00
You can run arbitrary .exe files in a web job, so theoretically I don't see why not. As to whether you ought to... :)
Bartosz Sypytkowski
@Horusiath
Sep 16 2016 06:40
@DamianReeves my CQRS example is slightly outdated. Regarding structs this won't help a lot (I think), because messages bo through untyped mailbox so structs will be boxed anyway.
regarding web jobs - it's possible, I know that problem may be with exposing enpoint in cluster scenarios
Shuffzord
@Shuffzord
Sep 16 2016 09:05
Hey guys, is there any room for some technical help in c# akka.net ?
Bart de Boer
@boekabart
Sep 16 2016 09:10
stackoverflow, of course, and this room.
Shuffzord
@Shuffzord
Sep 16 2016 09:20
I'm trying to setup nlog to be default logger for akka.net. I've included entries in config. loggers = ["Akka.Logger.NLog.NLogLogger, Akka.Logger.NLog"] with proper (i think) nlog entries but i still get buslogger as the default logger. How can i debug what config is used for system creation and what logger is picked?
Shuffzord
@Shuffzord
Sep 16 2016 09:34
To add more depth to the qeustion. I tried to create cluste rsystem with and without explicit config var section = (AkkaConfigurationSection)ConfigurationManager.GetSection("akka"); var system = ActorSystem.Create("webcrawler2"); ClusterSystem = ActorSystem.Create("webcrawler", section.AkkaConfig);
In both cases, cluster system in its 'settings' has loggers with 1 entry of nlog but the Logger of cluster system is set us buslogger
Bartosz Sypytkowski
@Horusiath
Sep 16 2016 11:02
@Shuffzord what exact path for loggers config did you use?
Shuffzord
@Shuffzord
Sep 16 2016 11:20

@Horusiath this is from my config file. Akka: <akka> <hocon> <![CDATA[ akka { loglevel = DEBUG loggers = ["Akka.Logger.NLog.NLogLogger, Akka.Logger.NLog"] log-config-on-start = on } ]]> </hocon> </akka>

And configsections deifinition

    <section name="akka" type="Akka.Configuration.Hocon.AkkaConfigurationSection, Akka" />
    <section name="nlog" type="NLog.Config.ConfigSectionHandler, NLog"/>
    <section name="entityFramework" type="System.Data.Entity.Internal.ConfigFile.EntityFrameworkSection, EntityFramework, Version=6.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089" requirePermission="false" />
    <!-- For more information on Entity Framework configuration, visit http://go.microsoft.com/fwlink/?LinkID=237468 -->
  </configSections>

And nlog section

<nlog xmlns="http://www.nlog-project.org/schemas/NLog.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" autoReload="true" throwExceptions="true" internalLogFile="C:\Temp\internal_log_file.txt" debug="true" internalLogLevel="Trace"> <targets> <target name="f1" xsi:type="File" fileName="Scrapper.logfile" layout="[${date}] [${threadid}] [${message}]" /> <target name="c" xsi:type="Console" layout="[${threadid}] [${message}]" /> </targets> <rules> <logger name="*" writeTo="f1" /> <logger name="*" writeTo="c" minlevel="DEBUG" /> </rules> </nlog>

tried to create actiorsystem in following ways
var section = (AkkaConfigurationSection)ConfigurationManager.GetSection("akka"); var system = ActorSystem.Create("webcrawler2"); ClusterSystem = ActorSystem.Create("webcrawler", section.AkkaConfig);
As i see, first version is basically what is done is original source code

Ohh and getting logger is private readonly ILoggingAdapter _logger = Context.GetLogger();
Bartosz Sypytkowski
@Horusiath
Sep 16 2016 11:23
you don't need to access ConfigurationManager explicitly - app.config content will be loaded automatically
Shuffzord
@Shuffzord
Sep 16 2016 11:23
thats what i read but since it wasnt working i tried to do it manually
no difference though
Bartosz Sypytkowski
@Horusiath
Sep 16 2016 11:34
I've create a fresh project with Akka.Logger.NLog and had no problems with plugin itself (more with nlog config section within app.config) - but at least this proves, that NLog is used indeed
Shuffzord
@Shuffzord
Sep 16 2016 11:39
i dont think it sthe akka problem anymore i think its something in nlog
i will try to check internal logs
thanks for help though
wdspider
@wdspider
Sep 16 2016 14:02
@Shuffzord after running into a similar issue a couple months back, I concluded that nlog likes to fail silently when it can't parse its config file. I ended up hard-coding the following InternalLogger settings, so that it wouldn't be quite so silent:
public class Program
{
    public static void Main(string[] args)
    {
        // Ensure NLog logs errors even if config file can't be parsed
        InternalLogger.LogLevel = LogLevel.Error;
        InternalLogger.LogToConsoleError = true;
        InternalLogger.LogToTrace = true;
        InternalLogger.LogFile = "nlog.errors";

        // Initialize topshelf service
        HostFactory.Run(x =>
        {
            x.SetServiceName("MyService");
            x.SetDisplayName("My Service");

            x.UseNLog();

            x.StartAutomaticallyDelayed();
            x.Service<MyService>();
        });
    }
}
Kevin Avignon
@Kavignon
Sep 16 2016 15:21
Hi. I have a quick question. On the akka-bootcamp repo, it mentions that the support for F# is accepted. What's currently lacking with the F# implementation compared to the C# one ? I'd like to be able to implement an event-based architecture with Akka.NET to create an rpg prototype. Am I getting over my head here doing it with F# ?
Bartosz Sypytkowski
@Horusiath
Sep 16 2016 15:41
@Kavignon default F# API doesn't allow you to describe some things that are necessary for features like cluster sharding or cluster singleton (Akkling API does however) - in general F# API is less used and less tested and you'll have less examples for it - usual thing in smaller communities.
Damian Reeves
@DamianReeves
Sep 16 2016 16:48
I'm looking to begin work on an Akka.Persistence plugin for Marten. The required serialization format is JSON, is there a way for me to have messaging use Wire and Persistence use JSON?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 16:52
@DamianReeves yep
some of our other plugins use JSON for persistence
I think the Sqlite one does
am I right about that @alexvaluyskiy ?
Vagif Abilov
@object
Sep 16 2016 16:54
@Kavignon while F# API is less used and lack few details we in our project find the missing pieces non-significant and are implementing the system in F# and Akka.NET. We do not use cluster though.
Damian Reeves
@DamianReeves
Sep 16 2016 16:55

cool. Is this something the user has to remember or can the persistence plugin enlist the correct serializer.

Meaning if change my serializer to wire and then I switch my persistence plugin to one requiring JSON, can I ensure I don't get wire formatted chunks sent to the persistence writer?

Damian Reeves
@DamianReeves
Sep 16 2016 17:00
Ok, I think I got it. Looked at the Redis provider and see a section on serialization:
akka.actor {
  serializers {
    redis = "Akka.Serialization.YourOwnSerializer, YourOwnSerializer"
  }
  serialization-bindings {
    "Akka.Persistence.Redis.Journal.JournalEntry, Akka.Persistence.Redis" = redis
    "Akka.Persistence.Redis.Snapshot.SnapshotEntry, Akka.Persistence.Redis" = redis
  }
}
Ronnie Overby
@ronnieoverby
Sep 16 2016 17:23
Does akka.net have a mechanism for rendezvous multiple async operations? Think Task.WhenAll
For example I'm processing a file. I might have many actors operating in parallel doing different things but all at the same time and I want to send a notification to a user's email address when I'm finished with all of the sub processes involved in processing the file.
I'm trying to figure out how to coordinate signaling that ALL of the actors involved have finished their part of the processing
Ronnie Overby
@ronnieoverby
Sep 16 2016 17:28
Of course I have ideas about how to build such a system, but I'm wondering if the AKKA has this figured out for me :)
Vagif Abilov
@object
Sep 16 2016 17:34
@ronnieoverby IMHO the actor models's way of achieving this is to implement the actor that needs to dispatch many parallel tasks and collect its results as a reactive accumulator: it sends messages to all worker actors at once and falls asleep. The worker actors sends back its results which are being accumulated by the dispatcher.
Ronnie Overby
@ronnieoverby
Sep 16 2016 17:36
I think it's a great opinion.
Alex Valuyskiy
@alexvaluyskiy
Sep 16 2016 17:39
All sql plugins use FindSerializerForType
Vagif Abilov
@object
Sep 16 2016 17:39
There is also a great post by @Horusiath on the philosophy of Akka: http://bartoszsypytkowski.com/dont-ask-tell-2/
Alex Achinfiev
@aachinfiev
Sep 16 2016 19:26

I am trying to play around with EventsByTag in Akka Persistence but running into an issue. I have defined config:

// In config
event-adapters = {
       security-tagger = "AkkaPersistenceTest.SecurityTagger, AkkaPersistenceTest"
}
event-adapter-bindings = {
        "AkkaPersistenceTest.SecurityEntity, AkkaPersistenceTest" = security-tagger
}

public class SecurityEntity
    {
        public Guid Id { get; set; } 
        public string Name { get; set; }

        public string Identifier { get; set; }
    }

public class SecurityTagger : IWriteEventAdapter
    {
        public SecurityTagger(ExtendedActorSystem system)   // For some reason it actually wants a constructor with actor system. 
        {
        }

        public string Manifest(object evt)
        {
            return "";
        }

        public object ToJournal(object evt)
        {
            var e = evt as SecurityEntity;
            if (e == null) return evt;
            Console.WriteLine("Tagging security {0} with {1}", e.Id, e.Identifier);
            return new Tagged(e, new[] { e.Identifier });
        }

I see it instantiate the Tagger but never calls the ToJournal and never records the actual tags on the persistence entry. I am using Akka.Persistence.Cassandra plugin based on the PR submitted by Constantin. Am I missing some configuration? I see the actual persistence working and event recorded, just without the tags. Thanks.

qwoz
@qwoz
Sep 16 2016 19:26
@Aaronontheweb interesting writeup of DigiOutsource. I noticed they're planning to use Service Fabric -- is there any documentation of how they will be setting this up? A (semi-) official starter solution demonstrating a basic 5 node akka.net cluster bootstrapped from Service Fabric would be very useful.
Aaron Stannard
@Aaronontheweb
Sep 16 2016 19:27
@qwoz second time I've had that question since publishing that this morning lol
I think I'm just going to bite the bullet and do something myself on SF soon
qwoz
@qwoz
Sep 16 2016 19:31
awesome... it's something I've been meaning to play around with myself, but other priorities always get in the way (geez... the business actually wants new features and bugs fixed?!?). And getting started using something from the Powers That Be™ that has proven itself in a production scenario would likely save hitting a lot of pitfalls along the way.
Alex Achinfiev
@aachinfiev
Sep 16 2016 19:40
@aachinfiev Update on my own question. I have to actually map the events rather than the state type which I was doing. Now I see event is being recorded and showing up in the eventsbytag1 table.
Chris Martin
@trbngr
Sep 16 2016 20:21
Hey guys. Anyone know of any open jobs? Scala,C#, frickin VB.. I don’t care ;)
Bartosz Sypytkowski
@Horusiath
Sep 16 2016 20:33
@DamianReeves newer versions of Akka.Persistence.PostgreSql plugin will have native support for json/jsonb data type
Damian Reeves
@DamianReeves
Sep 16 2016 20:35
that will be a great feature. @Horusiath do you have a recommendation for what persistence plugin is the best starting point for a new persistence plugin writer; I've been splitting between looking at MongoDb, Redis, and the EventStore plugin located here: https://github.com/vladkosarev/Akka.Persistence.EventStore
I would thing EventStore has more in common with Marten since they both natively implement event sourcing
But the EventStore plugin is 7 months old
Alex Achinfiev
@aachinfiev
Sep 16 2016 20:56
If I want to determine if a given entity already exists in the system using a given tag for it, is using EventsByTag stream an appropriate way to do so? Is there some sort of marker that tells me that there are no events and that means this entity doesn't exist, and reading one event would give me the latest event info that has been processed? Thanks.
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:17
I gotta put this damn Mono adventure to bed
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:18
Hi guys, does Context.SetReceiveTimeout(null) disable receving ReceiveTimeout messages?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:18
so close
@mmisztal1980 yep
that's how you cancel them
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:18
I seem to have a problem with it, probably some execution path where I don't set it...
what's up with Mono by the way?
8 tests shy
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:19
NIC
E
lol
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:19
trying to figure out if I can do anything about that
in one case it's just a race condition caused by limited degrees of parallelism, which can sometimes happen on our Windows builds too
the others I'm not so sure about
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:20
once the dust settles after that event on oct 3rd/4th I intend to have a look at Mono as well - there's a training session in town on Docker/Mesos - I'm trying to get in there
should open plenty of interesting options
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:20
yeah
I'm doing the same thing - want to get my service that I started blogging about on Petabridge up and running on Kubernetes or Mesos
maybe Docker Swarm
having the option to do that on Linux
is pretty amazing
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:22
precisely
and probably cheaper then using Windows
lol
btw, have you checked out cloud.docker.com? not 100% sure how that works, but it seems to integrate with docker swarm
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:33
not yet
that's the new proprietary Docker cloud right?
btw, update for you on your router issue: akkadotnet/akka.net#2266
Alex Achinfiev
@aachinfiev
Sep 16 2016 21:33
@Aaronontheweb I am actually trying to get to running on Kubernetes but we need to update some services to run on mono since we have some Windows dependencies. Hopefully get the in a next few weeks or so.
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:33
just figured it out a few seconds ago, I think
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:38
yep
oh nice!
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:39
99% sure that is the issue
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:39
btw, does Become affect Context.SetReceiveTimeout ?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:39
would be very easy to cause a race condition there
that would fuck up the routing table
really surprised this hasn't been an issue before
guess if people let the router warm up first it wouldn't have an issue
which is why the MNTR hasn't detected this before
@mmisztal1980 it doesn't
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:40
right-o
what about the nested Become issue?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:40
going to have to backlog that one for a later patch
delayed 1.1.2 for too long
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:41
I haer ya
hear*
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:41
been scarce on time this week
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:44
same :/ crazy times at work
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:44
@Silv3rcircl3 so these are the specs that are causing the most trouble with the Mono build... about 5 TCK specs that all report this error
Publisher Akka.Streams.Implementation.VirtualProcessor`1[System.Nullable`1[System.Int32]] did not drop reference to test subscriber after subscription cancellation
at Reactive.Streams.TCK.TestEnvironment.VerifyNoAsyncErrorsNoDelay () <0x414aad30 + 0x0009b> in <filename unknown>:0
at Reactive.Streams.TCK.PublisherVerification`1+<>c__DisplayClass58_0[T].<Required_spec313_cancelMustMakeThePublisherEventuallyDropAllReferencesToTheSubscriber>b__1 (IPublisher`1 publisher) <0x414cecb0 + 0x000eb> in <filename unknown>:0
at Reactive.Streams.TCK.PublisherVerification`1[T].ActivePublisherTest (Int64 elements, Boolean completionSignalRequired, System.Action`1 run) <0x414c06f0 + 0x00104> in <filename unknown>:0
at Reactive.Streams.TCK.PublisherVerification`1[T].Required_spec313_cancelMustMakeThePublisherEventuallyDropAllReferencesToTheSubscriber () <0x414cea70 + 0x001fb> in <filename unknown>:0
at Reactive.Streams.TCK.IdentityProcessorVerification`1[T].Required_spec313_cancelMustMakeThePublisherEventuallyDropAllReferencesToTheSubscriber () <0x4159ab60 + 0x0001b> in <filename unknown>:0
looks like there's something Mono-specific that affects cancellation support in the TCK
any ideas as to what the common element could be here? You can see the specific failures listed here (on the previous build, not the one currently running): http://petabridge-ci.cloudapp.net/viewLog.html?buildId=15948&tab=buildResultsDiv&buildTypeId=AkkaNet_LinuxMonoBuild&guest=1
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:48
btw, I had a questions regarding the crawler sample - on the web node, the /tasker router is a consistent-hash-router that talks to an ApiMaster on the other end - why did you pick this particular method? I seem to notice that the CommandProcessor that has an IActorRef to the router use the consensus algorithm that you typically employ in there , I was curious if that doesn't contradict consitent-hash-routing? the messages were supposed to end up on the same actor? At the end I've noticed that the messages sent from the CommandProcessor to the ApiMaster via the /tasker router, don't seem to implement any of the interfaces that consistent-hash-router uses - why? I'm a bit puzzled, because I wasn't able to piece the info together based on docs alone
correction IStartJobV1 does seem to contain the ConsistentHashKey so I guess that takes care of my 3rd question :D
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:55
the consistent hash router choice was simple
if someone tries to issue the same job twice
both job attempts go to the same node
so it's really easy to eliminate a duplicate that way
since we use the URL of the website to crawl as the hash key
Maciek Misztal
@mmisztal1980
Sep 16 2016 21:58
ahhh I see
Aaron Stannard
@Aaronontheweb
Sep 16 2016 21:59
@Silv3rcircl3 woot woot! just down to those 5 TCK specs
otherwise Mono builds pass!!!
(some race issues with things like the Dispatcher spec... trying to configure the Docker container to allow it to use up to 100% of the CPU so that doesn't happen often)
by default the Docker container should be able to do that anyway AFAIK but I'm not totally certain
but those are specs that rarely fail on the Windows builds
Marc Piechura
@marcpiechura
Sep 16 2016 22:00
sorry I'm busy right now, can't look into it
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:00
no problem
Marc Piechura
@marcpiechura
Sep 16 2016 22:00
the actual test is here
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:00
I'll look at it
maybe it's the GC stuff or so
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:01
I might just mark these specs as skipped for now - might need to add Mono builds to the Reactive.Streams repo first
verify that the TCK itself is Mono-compatible
which is a separate issue from Akka.Streams
makes sense, no?
Marc Piechura
@marcpiechura
Sep 16 2016 22:02
jup
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:03
k
that sounds like a reasonable compromise
Marc Piechura
@marcpiechura
Sep 16 2016 22:03
I also would say it's fine if this test only works on windows
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:03
yeah
truth of the matter is, we fixed all of the important stuff
that would have caused trouble on Mono
IPV6, Akka.IO, those expression compiler issues, the scheduler clean-up, etc
Marc Piechura
@marcpiechura
Sep 16 2016 22:04
if it releases the reference mono will also collect it even if it doesn't work with the spec
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:04
I'll get this PR cleaned up and ready for review
Marc Piechura
@marcpiechura
Sep 16 2016 22:04
that's the only thing it tests
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:04
yeah
those specs will still run on Windows
my test filter I wrote in the FAKE script
is platform-specific
Marc Piechura
@marcpiechura
Sep 16 2016 22:04
great, then we are fine ;)
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:04
stuff like the API diffing tool
can't run on Windows
err
Mono
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:22
nevermind, that wasn't the issue with the routers
that code was a bit convoluted but TL;DR; it's safe
Corneliu
@corneliutusnea
Sep 16 2016 22:35
guys, I'm still struggling to deploy Akka Cluster on an Azure App Service. I'm thinking either to try to write a proxy to make the cluster nodes talk via an LH to each other or write a transport to talk over HTTP/WebSockets with eachother. Both seem like crazy to achieve tasks. Thoughts?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:42
@corneliutusnea just so I don't sound like an idiot
Azure App Service
this is something that manages all of the hardware for you right
like you're no longer dealing with VMs directly?
Corneliu
@corneliutusnea
Sep 16 2016 22:46
yes, it's pure web app, seems that only HTTP/HTTPS services are enabled. They also scale up/down. I'm really sick of managing hardware so want a fully managed solution. I'm not worried about the cluster nodes finding themselves, I'm just trying to find a way for them to talk to each other
Corneliu
@corneliutusnea
Sep 16 2016 22:52
I'm thinking writing an Akka transport over WebSockets (instead of Helios) might be a way forward
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:52
@corneliutusnea if the environment is locked down
so TCP / UDP isn't an option
due to firewall whitelisting and loads of other fun stuff (like IIS being 100% unaware of any non-HTTP or WebSocket TCP traffic)
Corneliu
@corneliutusnea
Sep 16 2016 22:53
no, it's not an option, unless I want to move to something like ASE (Application Service Environmnet) but that's a fully managed environment .. $20K up
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:53
in that case
WebSockets is an idea
can probably build something on top of a SignalR persistent connection
some of my compadres in the mobile gaming space have done that for similar reasons
not for Akka.NET
hand-rolled stuff
but it's achievable
other possibility if you're open to it
checked out Azure Container Service yet?
that might give you a best of both worlds option
you don't really run the hardware
just deploy containers onto it
and the hardware itself is managed by Azure
supposed to be Mesos-compatible too
Corneliu
@corneliutusnea
Sep 16 2016 22:56
hm, didn't really mentally attach to the whole container thing .. might need to try it
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:56
Mono support is coming imminently
which makes that possible today
Corneliu
@corneliutusnea
Sep 16 2016 22:57
don't really care about mono :D
Bart de Boer
@boekabart
Sep 16 2016 22:57
You should if you want to run (linux) containers :)
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:57
Mono is necessary because fat-ass .NET can't be Dockerized
stuck with .NET Core, which is going to take a bit longer
or Mono
Corneliu
@corneliutusnea
Sep 16 2016 22:57
I really see no value proposition for Mono unless you REALLY badly want to deploy on Linux .. given we have zero expertise in Linux in my team .. that's not on the cards
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:58
the argument there isn't really for Linux
it's for Docker
problem is that Docker on Windows Server is still in beta
and pretty immature
Corneliu
@corneliutusnea
Sep 16 2016 22:58
I'd like to stick to the app service thing or try Windows Nano just because it's to bare metal fast
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:58
give Windows Nano a try
I haven't done much with that yet
if there's a Windows Nano docker image out there that supports a full .NET profile install
that'd give you what you need
Corneliu
@corneliutusnea
Sep 16 2016 22:59
yeah, I'm playing with it, looks nice, just limited .Net on it still but I think that's the future .. power/$ ratio is crazy ..
Aaron Stannard
@Aaronontheweb
Sep 16 2016 22:59
for Azure Container Service
Corneliu
@corneliutusnea
Sep 16 2016 23:00
so if I want to write my own transport for Akka .. where do I start? HeliosTransport class as an example?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:01
HeliosTransport
or you can look at the 1.5 branch
and check out the network stream transport
Corneliu
@corneliutusnea
Sep 16 2016 23:02
"network stream transport" ? can't find such a class
this is what will end up replacing Helios eventually
.NET Core-friendly and faster than Helios
also supports TLS
Corneliu
@corneliutusnea
Sep 16 2016 23:04
I plan to go live in about 2-5 weeks .. should I attempt to jump on 1.5 ?
or just use that as an example?
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:04
use that as an example
we're still waiting on Wire to be finished
for 1.5
well, not "finished"
but "production-certified"
I have work to do on Wire too - haven't added automated CI for it yet
Corneliu
@corneliutusnea
Sep 16 2016 23:05
any ETA? I might be able to wait ~6-7 weeks
don't want to develop and then throw it away
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:07
not sure
once we get 1.1.2 out the door I'll have more time to spend on it
fixing both of the bugs blocking it now
Corneliu
@corneliutusnea
Sep 16 2016 23:10
ok, thanks, I might look into 1.5 then and maybe attempt to build a WebSockets transport
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:14
all green on Mono certification!!!!
woohooo!
Corneliu
@corneliutusnea
Sep 16 2016 23:14
:D
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:14
going to rebase everything and update my PR shortly
almost have this cluster router bug figured out too
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:19
@mmisztal1980 fixed it!
man, it was the dumbest thing
we weren't reading in the correct defaults inside the ClusterDeployer
that's all
all of the actual Akka.Cluster code worked fine
just selected the wrong HOCON key
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:44
#2310 will fix it
Aaron Stannard
@Aaronontheweb
Sep 16 2016 23:51
and completed Mono support: akkadotnet/akka.net#2311
Corneliu
@corneliutusnea
Sep 16 2016 23:57
@Aaronontheweb taking a stab in the dark, do you think Akka 1.5 will be ready by December-ish?