These are chat archives for akkadotnet/akka.net

20th
Sep 2017
Saepul Ramdani
@blackclavus
Sep 20 2017 05:00
[ERROR][9/20/17 4:50:41 AM][Thread 0007][akka://mysystem/system/IO-UDP-FF/$a] Provided length of array to copy [265] doesn't fit array length [64000] and offset [63744].
Parameter name: count
Cause: System.ArgumentException: Provided length of array to copy [265] doesn't fit array length [64000] and offset [63744].
Parameter name: count
   at Akka.IO.ByteString.FromBytes(Byte[] array, Int32 offset, Int32 count)
   at Akka.IO.UdpListener.DoReceive(SocketAsyncEventArgs e, IActorRef handler)
   at Akka.IO.UdpListener.ReadHandlers(Object message)
   at Akka.IO.UdpListener.<PreStart>b__11_0(Object m)
   at Akka.Actor.ActorBase.AroundReceive(Receive receive, Object message)
   at Akka.Actor.ActorCell.ReceiveMessage(Object message)
   at Akka.Actor.ActorCell.Invoke(Envelope envelope)
Just got this error, I'm still investigating why this is happen
(I'm using Akka 1.3.1 with dotnet core 2.0)
Alex Michel
@amichel
Sep 20 2017 11:09
In ReceivePersistentActor, what is the best way to intercept all commands? I want to asynchronously log all incoming commands before they are passed to registered handlers for specific message types.
Zetanova
@Zetanova
Sep 20 2017 12:15
@amichel override AroundReceive
there u can make a pre-receive, post-receive pattern or just call a custom-generic receive handler
Aaron Stannard
@Aaronontheweb
Sep 20 2017 13:57
@uatec https for getakka.net now works
@blackclavus weird - looks like the code used for working with the internal buffers inside our Akka.IO UDP implementation has a bug there. Could you please log an issue along with some reproduction steps (i.e. sending large sequences? lots of small ones? etc)
Ismael Hamed
@ismaelhamed
Sep 20 2017 14:05
Unable to cast object of type 'Surrogate' to type 'Akka.Actor.ActorPath'.
Hyperion 0.9.2
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:05
@ismaelhamed I think I logged that one here...
Ismael Hamed
@ismaelhamed
Sep 20 2017 14:05
ok
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:06
akkadotnet/Hyperion#75
does that look right to you ?
Ismael Hamed
@ismaelhamed
Sep 20 2017 14:06
let me try your work around
ok, we've now moved to an Object reference not set to an instance of an object.
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:11
ugh
what type of message is it you're trying to send over the wire?
Saepul Ramdani
@blackclavus
Sep 20 2017 14:11
@Aaronontheweb I'm still working on sample code, because I'm not sure how to fully reproduce that bug. I'll log issue asap
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:12
thanks!
Ismael Hamed
@ismaelhamed
Sep 20 2017 14:12
well, don't you judge me just yet... but this is a persistent snapshot
It happens after a bunch of snapshots, it gets to a point where it won't deserialize it anymore
mrrd
@mrrd
Sep 20 2017 14:32
i am currently trying to get remote communication with an akka cluster using pbm client 0.2.2 (petabridge.cmd), it works locally but not remotely, remotely it attempts to connect but just hangs without "successfully connection to "... being displayed, is there are known issue in this version of petabridge.cmd and remote connections? I have tested using telnet on the remote machine and that connects successfully without issue so it would not appear to be a firewall issue?? Is this something that would be fixed once we find the time to upgrade the cluster to 1.3.1 (and the relevant version of petabridge.cmd) ?
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:33
@mrrd we've pushed a 1.3.1-compatible version of Petabridge.Cmd as of last week
v0.3.0
the v0.2.2 versions should still work fine in .NET 4.* apps but won't work at all on .NET Core
you may want to check the binding address you're using in the PBM host hocon settings
mrrd
@mrrd
Sep 20 2017 14:34
our cluster is currently running akka 1.2.3
mrrd
@mrrd
Sep 20 2017 14:34
yes checked that, and also checked the output of netstat to verify
its definately binding to 0.0.0.0
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:35
hmmm
and it just hangs eh
and you can produce this 100% of the time?
mrrd
@mrrd
Sep 20 2017 14:37
the strange thing is it works locally on the server perfectly, but we do not want to have to keep RDP'ing in to use pbm, and its not a firewall issue as telnet from the same remote machine(s) can establish a connection if using telnet (I always try that out to check for firewall issues)
yes 100% of the time from the remote machine
its a strange one
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:38
weird
the thing that makes me wince at that is the pbm client doesn't even time out, right?
normally it'd show a "could not connect" error after 5 seconds
mrrd
@mrrd
Sep 20 2017 14:39
telnet works remotely, pbm client does not
exactly
it doesnt even show that
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:39
ok, well that gives me some clues
as to where to look
Alex Michel
@amichel
Sep 20 2017 14:39
@Zetanova thanks. wish I also had "AroundSend" api, to optionally log the outgoing messages as well.
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:39
sure sounds like it's an issue with the pbm client
Bartosz Sypytkowski
@Horusiath
Sep 20 2017 14:39
@Aaronontheweb try to start with telneting akka.io tcp server locally - I've got an issue there, but haven't time to verify
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:40
@mrrd that's a good idea cc @Horusiath
mrrd
@mrrd
Sep 20 2017 14:40
cool, its v0.2.2 if that helps, was wondering if anyone else had experienced that and if it was a known issue that will be resolved once we get the chance to upgrade everything to 1.3.1
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:40
but this is also with the 1.2.* Akka.IO bits
not the new ones in Akka 1.3.1
@mrrd would you mind logging a brief issue here? https://github.com/petabridge/petabridge.cmd-issues
I can take a look at it this week
mrrd
@mrrd
Sep 20 2017 14:41
sure @Aaronontheweb
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:41
did a maintenance release of pbm yesterday to fix a bug in the new 0.3.0 version
mrrd
@mrrd
Sep 20 2017 14:41
thanks
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:44
I'll take a look and see if I can produce a fix
if I can make it retroactive to the v0.2.* version I'll do that
depends on whether or not Chocolatey is going to be nice to me :p
Bartosz Sypytkowski
@Horusiath
Sep 20 2017 14:45
@james-world if you use SQL journal (currently SqlServer or Sqlite), you can define a "default" serializer just for that journal - i.e. for default SqlServer plugin it would be akka.persistence.sql-server.journal.serializer = json
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:50
@ismaelhamed hmmmm
that's interesting
so it's using those preserving those references.. but then they "degrade" after a while?
mrrd
@mrrd
Sep 20 2017 14:53
@Aaronontheweb thanks, I have raise the issue as #19, if you can get to the bottom of it it would be a great help, or even if verifying if its something that will be fixed by upgrading our cluster to 1.3.1 so that we can increase the priority of our planned upgrade :)
Aaron Stannard
@Aaronontheweb
Sep 20 2017 14:53
@mrrd Petabridge.Cmd changed pretty significantly in v0.3.0
the core functionality is still the same
but Akka.IO, which we depend on internally, now uses SocketAsyncEventArgs and that was a large change
and we rewrote all of the serialization to use Protobuf instead of Hyperion so we could achieve x-plat compatibility between .NET Core and .NET 4.* apps with a single pbm client
so it's tough for me to say if upgrading will help
I'll investigate by going back in time first
and seeing how the v0.2.2 stuff holds up and if I can reproduce this issue or not
mrrd
@mrrd
Sep 20 2017 14:59
yes that one thing that is delaying our upgrade to 1.3.1 we are planning to switch to use Protobuf serialization from Hyperion so we are just wanting to be absolutely certain everything is not going to break horrifically when doing so and ensure we get all the settings right as we will be wanting to evaluate .net core connectivity too as we have some .net core that it would be useful be able to directly join the cluster, hopefully you'll be able to reproduce what I am seeing and at the very least confirm if 1.3.1 will be how we resolve it :+1:
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:02
the protobuf serialization stuff takes a little while to set up
but a big fan of it thus far
for my own application-specific messages
mrrd
@mrrd
Sep 20 2017 15:06
cool thanks for your help @Aaronontheweb
Vagif Abilov
@object
Sep 20 2017 15:10
Trying to upgrade to Akka 1.3.1. TypeInitializationException on start:
"The type initializer for 'DotNetty.Transport.Channels.DefaultChannelId' threw an exception."
Bartosz Sypytkowski
@Horusiath
Sep 20 2017 15:10
@object any more info?
Vagif Abilov
@object
Sep 20 2017 15:12
This is what happenned immediately after upgrading from Akka 1.2.3/Akkling 0.6.1 to Akka 1.3.1/Akkling 0.7. I can check where it's happening.
Failed to bind to "0.0.0.0:1954"; shutting down DotNetty transport.
This is what I have in HOCON
<akka.remote>
  <hocon>
    <![CDATA[
      akka.remote.dot-netty.tcp {
        port = 1954
        hostname = 0.0.0.0
        public-hostname = localhost
      }
  ]]>
  </hocon>
</akka.remote>
Prior to upgrade this worked fine.
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:15
weird
that should still work fine
what version of DotNetty are you running?
Vagif Abilov
@object
Sep 20 2017 15:16
0.4.6
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:16
and IIRC, that's part of the 1.3.1 upgrade
yeah we were on v0.4.3 of DotNetty in 1.2.3
so it might be a change they made there to that code responsible for generating the DefaultChannelId
could you try starting up with Just My Code disabled?
and see if it throws an exception down in there?
and what environment is this on?
Vagif Abilov
@object
Sep 20 2017 15:17
Just a sec
Windows 8.1
Ronny Carlansson
@lessismore1
Sep 20 2017 15:20
I have action handlers - Command<> - defined in a derived class. I want to execute them before default behaviour in base class. How to do that?
Vagif Abilov
@object
Sep 20 2017 15:20

Exception thrown: 'System.IO.FileNotFoundException' in DotNetty.Common.dll

Additional information: Could not load file or assembly 'Microsoft.Extensions.Logging.Abstractions, Version=1.1.2.0, Culture=neutral, PublicKeyToken=adb9793829ddae60' or one of its dependencies. The system cannot find the file specified.

And then the other exception is thrown.
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:21
ah yeah that would do it
Vagif Abilov
@object
Sep 20 2017 15:21
But where does 'Microsoft.Extensions.Logging.Abstractions come from?
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:21
what version of that package is installed?
it's a dependency of DotNetty
that's the new piece of centralized middleware all MSFT products use for logging
Vagif Abilov
@object
Sep 20 2017 15:21
DotNetty.Common (0.4.6)
  Microsoft.Extensions.Logging (>= 1.1.1)
From paket.lock
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:22
could you check the packages folder and make sure that's in there?
Vagif Abilov
@object
Sep 20 2017 15:22
No it's not there
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:22
weird
I'm not that familiar with Paket - any reason why it wouldn't install it?
Vagif Abilov
@object
Sep 20 2017 15:23
Should I include that file in all bin directories where DotNetty resides?
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:23
I mean, Paket should just install it like any other NuGet dependency
(sorry for the constant typos, getting used to a new keyboard....)
Vagif Abilov
@object
Sep 20 2017 15:24
But Paket only installs in the project bin folder files explicitly listed in paket.references (unless of course the asembly drags it as dependency)
So paket.references for all executable projects should list Microsoft.Extensions.Logging, I suppose.
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:26
DotNetty.Common does list it as a dependency though.... https://www.nuget.org/packages/DotNetty.Common/
might be a Paket bug of some kind maybe
but yeah, should definitely make sure that's in the output folder
Vagif Abilov
@object
Sep 20 2017 15:28
Making a change...
But are you sure that NuGet packages listed as dependencies are always installed in the output folder? AFAIK if the assembly doesn't reference dependent lib, it will need to be explicitly referenced by the project (or listed in paket.references if Paket is used). Isnt' it so?
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:44
these days with how everything has changed with the inclusion of .NET Standard and MSBuild 15 I'm not so sure any more
Vagif Abilov
@object
Sep 20 2017 15:44
Wow. Even listing it in paket.references didn't result in the library copied in the output dir.
I am beginngin to suspect the version of Microsoft.Extensions.Logging. I didn't have any constrained on version, so the latest on (2.0) was installed. Trying version 1.
Vagif Abilov
@object
Sep 20 2017 15:51
Yepp, I set a constraint on Microsoft.Extensions.Logging version ~1 and now I am getting missing Microsoft.Extensions.Logging.Abstractions
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:52
oh wow... so the issue is that the .NET Standard 2.0 version was being installed?
Vagif Abilov
@object
Sep 20 2017 15:52
Exactly. Unless I constrained the version to 1.x
Aaron Stannard
@Aaronontheweb
Sep 20 2017 15:52
ughhhhhhh
so the DotNetty guys will need to fix that... or we'll need to start adding explicit upper bounds on the versions of the packages we use
we can't solve that particular one though since that's one of DotNetty's dependencies
but man, what a pain in the ass
Joshua Garnett
@joshgarnett
Sep 20 2017 15:54
I ended up having to flag CS1701 as nowarn due to different nuget packages wanting different versions of things. We run warns as errors and that one was causing headaches.
Vagif Abilov
@object
Sep 20 2017 15:54
Yes this is purely DotNetty stuff. But makes things even more painful because I wouldn't suspect that.
Vagif Abilov
@object
Sep 20 2017 16:03
Finally worked after adding to paket.dependencies a line
nuget Microsoft.Extensions.Logging.Abstractions ~> 1
and listing Microsoft.Extensions.Logging.Abstractions in paket.references for the application project.
Anders Storhaug
@andersstorhaug
Sep 20 2017 17:21
anyone consider using HOCON for app-level config?
Anders Storhaug
@andersstorhaug
Sep 20 2017 18:15
more specifically HOCON for non-Akka config, but still bootstrap-level stuff. I'm kinda considering it, wondering if any others have done it
Joshua Garnett
@joshgarnett
Sep 20 2017 18:16
We did for my Scala project, I’ve haven’t gotten far enough down my current project path to decide if I want to do the same for C#
Anders Storhaug
@andersstorhaug
Sep 20 2017 18:20
Been experimenting with it, and like it quite a bit so far. Dealing with somewhat complex config. Not aware atm of other config formats with same features in spec
Mark Gerrits
@skinny
Sep 20 2017 19:04
Hi guys, I'm very new to Akka.net but I'm trying to read up and experiment as much as I can to get up to speed. I'm currently trying to embed Akka into an existing application to see what it can do with already existing business logic. One apparent problem I'm bumping into is that DI seems sort of ackward (confirmed by some googling too).
I have an actor with the following static Props method : public static Props Props(Relation relation) { return Akka.Actor.Props.Create(() => new ProcessMessageActor(relation)); }
However the ProcessMessageActor also needs an existing class injected into the constructor such as : public ProcessMessageActor(Relation relation, IMyOtherService service)
I already installed the DI package (SimpleInjector) which works for constructing stuff with Context.DI()... but what's the best way to deal with this scenario ?
Roman Golenok
@shersh
Sep 20 2017 19:38
guys, is there any tools for managing akka actors? Something like Observer from Erlang. I want to see which actors exists and in which node, etc
Deniz İrgin
@Blind-Striker
Sep 20 2017 19:39
@shersh There is a tool. We're currently using it, let me find it
Roman Golenok
@shersh
Sep 20 2017 19:40
@Blind-Striker Thanks!
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:44

Hello, hello. I'm looking for an official answer to something I'm seeing, to guide me whether I'm, doing it wrong, or it just doesn't work how I think it should. If I start up a cluster with a dedicated seed node and two others, everything works fine no matter what order I start them in, and messages flow between nodes, cluster gossip bounces around etc. If either of the worker nodes goes down and is restarted, it joins back into the cluster and all is good.

My confusion is with what happens if the seed goes down. When it comes back up, it seems to join the cluster again, but gossip doesn't look right - there's none arriving at the seed, but the other nodes notice it's there. At this point, all other nodes need to be restarted to have the cluster back to normal function, with gossip hitting all nodes etc.

I thought the seed nodes in config were just like an initial population of a member list, but it seems they are more than that. I know all nodes should ideally see the same initial seed on start-up to avoid split brain, but why is the cluster not resilient to the seed going down and rejoining?

It looks like there must always be at least one seed node in the cluster, as I don't see the problem if I have two and kill/restart one at a time.

What am I missing?

Thanks, and sorry for the post length
Deniz İrgin
@Blind-Striker
Sep 20 2017 19:46
@shersh well it doesnt show which actors exists in which node, only shows nodes in cluster and give you ability to send leave or shutdown order to nodes. But i think with small modifications you can do it.
Joshua Garnett
@joshgarnett
Sep 20 2017 19:46
So I think the challenge is that your seed nodes will join each other, but not the cluster if they only know of each other
So you always do need at least one seed node running to recover properly
I try to run 3 seed nodes and treat anything more than 1 seed node loss as a total cluster failure.
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:48
@joshgarnett that doesn't make sense. I can start the nodes in any order initially and the workers just keep looking for the seed until it comes up. I thought they would do the same if it goes down, and when they find it again, re-share cluster info
Joshua Garnett
@joshgarnett
Sep 20 2017 19:49
Here let me type up a scenario
Ahh I see what you are sayin
No, once a node is evicted from the cluster, it will not try to reconnect to it after it is quarantined
Scenario:

Seed Nodes A, B, C
Worker Nodes D, E, F

Start the seed nodes
Start the worker nodes

We now have a cluster of A, B, C, D, E, F

Assuming all seed nodes go down, we have a cluster of D, E, F

Seed nodes come back up, we have two clusters,  A, B, C and D, E, F

If three more worker nodes come up (G, H, I), we have two clusters A, B, C, G, H, I and D, E, F
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:52
I'm seeing the worker nodes be reconnected when coming back up, with messages about them being the same node at the same address
will D,E,F not keep looking for the defined seeds to come back, or is the seed-nodes config only for starup?
Joshua Garnett
@joshgarnett
Sep 20 2017 19:53
They will not, it’s only on startup
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:53
ok, that's my issue then :)
Joshua Garnett
@joshgarnett
Sep 20 2017 19:53
That said, you can manually rejoin them to the cluster D, E, F
But it won’t happen automatically
There is the petabridge tool that’ll help with some of the scripting there
That said, yeah you usually don’t want your seed nodes to go down
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:55
to rejoin them, i'd need to code the workers to look for the seeds from config manually, I guess
Deniz İrgin
@Blind-Striker
Sep 20 2017 19:55
But i thought if one the nodes become unreachable its stays in cluster until some one or some strategy mark them as down.
And they could re-join cluster again
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:56
@Blind-Striker that's what i'm seeing with the workers. the're re-activated when coming back up
Joshua Garnett
@joshgarnett
Sep 20 2017 19:56
Depends on your strategy, you can setup autodown (dangerous as you can get split brains) or use a custom solution to handle downing safely
Each ActorSystem in the cluster comes up with a different nonce, so even if your ip:port are the same, its treated as a new node when it comes up
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:58
that wold explain the message "new incarnation of xyz" I saw
Joshua Garnett
@joshgarnett
Sep 20 2017 19:58
Yeah the gossip protocol is very basic at its core, you still have to do some heavy lifting on your side
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 19:59
they can be disconnected and reconnect (network issue), but going down and restarting is a new instance, not the same reconnecting, right?
Joshua Garnett
@joshgarnett
Sep 20 2017 19:59
My co-worker is working on an open source solution for split brain resolution at https://github.com/zynga/Zynga.Akka.Cluster
Correct, transient issues can be recoved, but an app restart should be a new node
RobPurcellCgl
@RobPurcellCgl
Sep 20 2017 20:00
this is something it would be nice to have some targeted examples of. Cluster node downing strategies, handling network partition etc. Best practice type stuff
thanks, i'll have a look at that
Aaron Stannard
@Aaronontheweb
Sep 20 2017 20:58
@shersh @Blind-Striker https://cmd.petabridge.com/
the actor hierarchy command
and it can be used to query actors on every node
Roman Golenok
@shersh
Sep 20 2017 21:03
@Aaronontheweb !!! thanks!
Aaron Stannard
@Aaronontheweb
Sep 20 2017 21:03
going to get a UI based version of this thing out the door later this year
brushing up on my TypeScript :p
but I'm fond of the command line
it's been helpful in debugging some issues with Akka.Cluster in the past
Joshua Garnett
@joshgarnett
Sep 20 2017 21:05
@Aaronontheweb what’s your poison of choice when it comes to JS frameworks?
Aaron Stannard
@Aaronontheweb
Sep 20 2017 21:05
I'm tempted to keep it old school
and go with KnockoutJs
had lots of experience working with Knockout and SignalR
and charting frameworks
they all go together great
I'm aghast to these frameworks that try to impose their own MVC stuff on top of my server-side code that... already does MVC
and I like having things like hyperlinks just work
kind of just want my JS to stfu, sit in the corner, and work properly on Internet Explorer
not try to sell me on buying a free trade all-organic Sumatran single origin pour over coffee
Joshua Garnett
@joshgarnett
Sep 20 2017 21:08
Oh man, dating yourself there :P
Can’t really be a JS dev unless you are always on the latest and greatest framework
Aaron Stannard
@Aaronontheweb
Sep 20 2017 21:10
I have never seen a person more panicked in my life than a JavaScript developer hearing about a new framework for the very first time
Joshua Garnett
@joshgarnett
Sep 20 2017 21:12
lol
Deniz İrgin
@Blind-Striker
Sep 20 2017 21:13
:D
Joshua Garnett
@joshgarnett
Sep 20 2017 21:13
preact is on my radar to check next
Deniz İrgin
@Blind-Striker
Sep 20 2017 21:14
I think Web Assembly gonna solve everything
Aaron Stannard
@Aaronontheweb
Sep 20 2017 21:14
I think my next software company is just going to automate away front-end development in its entirety, mostly out of spite for JavaScript developers specifically
Joshua Garnett
@joshgarnett
Sep 20 2017 21:15
Good luck
Aaron Stannard
@Aaronontheweb
Sep 20 2017 21:15
Silicon Valley will be the new rust belt
reduced to inventing Juicero 3.0 or whatever
I'm just kidding
I think WebAssembly will be cool
Joshua Garnett
@joshgarnett
Sep 20 2017 21:16
It’s going to do some cool things for WebGL gaming
Deniz İrgin
@Blind-Striker
Sep 20 2017 21:19
There is a project called Seif by Douglas Crockford. I know its not gonna be next big or be supported by the community. But its criticize current Web Architecture and suggest a new way of thinking.
Wow. I did'n know that the video thumbnail would be that big. Sorry :(
Aaron Stannard
@Aaronontheweb
Sep 20 2017 21:21
it's all good man
Gitter just wants you to know it REALLY loves you video
Augustine Gyawu Adjei
@austinejei
Sep 20 2017 21:59
lol
Andrew Spiering
@wackoisgod
Sep 20 2017 22:57
@object I tried putting in a PR to get DotNetty to upgrade their protobuf version and they pretty much said no :( They expect people to just maintain their own versions of it with the updated libs
and I don’t really have a good objection against that outside it being annoying
Johnny Z
@StormHub
Sep 20 2017 22:58
@wackoisgod what was the problem with the version in akka
rebinding cannot solve it?
Andrew Spiering
@wackoisgod
Sep 20 2017 22:59
I mean it can yes but is the resolution just going to be well we will never upgrade and people will see this warning and not really know what it means ? We ignored the warning and just assume stuff will work.
Johnny Z
@StormHub
Sep 20 2017 23:01
@wackoisgod dot netty stick to versions as low as it can possibly support
well, upgrade to the latest and ignore the warning :D
Andrew Spiering
@wackoisgod
Sep 20 2017 23:06
@StormHub sure
Johnny Z
@StormHub
Sep 20 2017 23:07
its a big annoying but does not do any harm :)
*bit
Andrew Spiering
@wackoisgod
Sep 20 2017 23:31
:thumbsup:
Aaron Stannard
@Aaronontheweb
Sep 20 2017 23:32
managing common versions of stuff across large projects is among my less favorite things to do
each time the ASP.NET team comes up with some sort of new universal common abstraction or whatever I feel my stomach acidity increase
Johnny Z
@StormHub
Sep 20 2017 23:32
lol
Aaron Stannard
@Aaronontheweb
Sep 20 2017 23:33
"WE'VE FINALLY DONE IT - WE HAVE A UNIVERSAL WAY OF HOW TO DO LOGGING. ALL WE NEED ARE THESE OTHER 10000 .NET OSS PROJECTS TO ADOPT IT"
"what happens if two popular OSS projects people want to use together use different, incompatible versions of this universal interface?"
"that will never happen, trust me."
6 months later INTRODUCING .NET CORE 2.1
harkens back to the JSON.NET assembly version fights of yore
Augustine Gyawu Adjei
@austinejei
Sep 20 2017 23:50
hello guys
i'm getting this error "Connection was reset by the remote peer. Channel "[::ffff:127.0.0.1]:61647"->"[::ffff:127.0.0.1]:1001"" when doing remoting
any help?
used to work before i upgraded to 1.3.1
Aaron Stannard
@Aaronontheweb
Sep 20 2017 23:53
that code should still work
didn't make any major changes there other than the DotNetty version upgrade
and some of the serialization stuff we added
that error message indicates it's a socket issue though
Augustine Gyawu Adjei
@austinejei
Sep 20 2017 23:53
i'm quite surprised tho
Aaron Stannard
@Aaronontheweb
Sep 20 2017 23:53
not Akka.Remote
means you connected successfully but something at the DotNetty / socket level aborted
Augustine Gyawu Adjei
@austinejei
Sep 20 2017 23:54
lemme check again
Aaron Stannard
@Aaronontheweb
Sep 20 2017 23:54
what's the error message on the remote node?
the one you're trying to connect to