Akka.NET

Canonical actor model implementation for .NET with local + distributed actors in C# and F#. Apache 2.0 licensed and free to use.

[Akka.NET @AkkaDotNET]

A fixed https://t.co/rEWXrRRKPZ CVE offers a useful design lesson: keep Akka.Remote on private networks, and use mTLS when traffic crosses trust boundaries. Add hostname validation or pinning when the threat model requires it.

https://t.co/1URJVHiyST

[Aaron Stannard @Aaronontheweb]

So @AkkaDotNET v1.6 is already an order-of-magnitude improvement for networking and serialization performance over v1.5, but the next beta of it is about to be even faster:

Looking at about a ~20% throughput improvement across the board and significant reductions in CPU / mem resources per message.

All from simple mechanical fixes plus some experimental IActorRef compression on the receiving side of the wire.

[Akka.NET @AkkaDotNET]

Kafka consumers must coordinate in-flight work, offset commits, and partition revocation. Akka.Streams.Kafka layers stream backpressure and committable offsets over Confluent.Kafka so those concerns stay in one pipeline.

https://t.co/SW5HBi4f8h

[Akka.NET @AkkaDotNET]

“Unreachable” does not mean “removed” in https://t.co/rEWXrRRKPZ. A busy process or short network partition may recover. Give partitions time to heal, then let a split-brain strategy decide when a member must be downed.

https://t.co/MWm2bLKgCc

[Aaron Stannard @Aaronontheweb]

Had a customer try out @AkkaDotNET v1.6 in their high-performance app today and report several different performance / code-generated serializer issues their agents found while exercising it.

The rate at which we can improve software now is just incredible

[Akka.NET @AkkaDotNET]

Kubernetes rollouts can trigger repeated shard rebalances. Petabridge recommends Deployments with `maxSurge: 1` and `maxUnavailable: 0` for new https://t.co/rEWXrRRKPZ projects, bringing up a replacement before removing the old pod.

https://t.co/XaBYcqBs6Y

[Akka.NET @AkkaDotNET]

With Akka.Cluster.Sharding, callers send to an entity ID instead of tracking a node address. Sharding locates or starts that entity and can redistribute its shard as cluster membership changes.

https://t.co/59rblcRHWA

[Akka.NET @AkkaDotNET]

An https://t.co/rEWXrRRKPZ actor can use `Become` to change which messages it handles in each state. Pair it with `IStash` to defer requests that aren’t valid yet, then unstash them when the actor returns to a ready behavior.

https://t.co/LoQMW8Ydic

[Akka.NET @AkkaDotNET]

https://t.co/rEWXrRRKPZ 1.5.73 is out. Important fixes for cluster extension startup and Cluster.Sharding. Plan your upgrade:

https://t.co/ASvwLVfkMS

[Akka.NET @AkkaDotNET]

https://t.co/hMgVjNMrv1 1.5.71-beta1 moves Quartz 3 to 4. It's .NET 10 only and breaking. This is a prototype — we want to validate the migration path before calling it production-ready. Try it and tell us what breaks.

https://t.co/9puGt1NcO7

[Akka.NET @AkkaDotNET]

Well, a beta of it at least 😘

[Akka.NET @AkkaDotNET]

v1.6 is out and ready for some testing

[Aaron Stannard @Aaronontheweb]

Going to do a for real post on this soon, but late last night we shipped @AkkaDotNET v1.6-beta1

This includes:

- Full AOT support for core Akka, Hosting, Streams
- Source-generated serialization + System.Memory-powered primitives for it + all I/O
- Artery.Tcp - need to re-benchmark it since Iast posted about it but TL;DR; it can do 1.6m msg/s over a single connection, and these are REAL message payloads, not byte arrays or strings.

I've already validated the AOT support in a live application that consumes Akka + Akka.Streams - works great so far. Going to deploy that later today.

[Aaron Stannard @Aaronontheweb]

Going to do a for real post on this soon, but late last night we shipped @AkkaDotNET v1.6-beta1

This includes:

- Full AOT support for core Akka, Hosting, Streams
- Source-generated serialization + System.Memory-powered primitives for it + all I/O
- Artery.Tcp - need to re-benchmark it since Iast posted about it but TL;DR; it can do 1.6m msg/s over a single connection, and these are REAL message payloads, not byte arrays or strings.

I've already validated the AOT support in a live application that consumes Akka + Akka.Streams - works great so far. Going to deploy that later today.

[Akka.NET @AkkaDotNET]

A read-only property doesn’t make an https://t.co/rEWXrRRKPZ message immutable if it still points to a mutable collection. Copy mutable state into a new array or immutable structure before sending it. This prevents cross-actor side effects.

https://t.co/lUStBn2j1V

[Akka.NET @AkkaDotNET]

If you use https://t.co/916SOQ1bBk, upgrade to 1.5.71 now.

It fixes two snapshot-store bugs that could use the wrong sequence number during recovery and potentially cause missing data.

Release notes: https://t.co/cv7Nrmg6f4

[Jaime @jaimefebres]

It's your time to shine @AkkaDotNET

[Petabridge @petabridge]

New live https://t.co/bwaZZvEppl training dates are available for Q4.

Join instructor-led sessions on Akka.Streams, Akka.Cluster, and end-to-end application design.

See the full calendar and reserve your spot:
https://t.co/AizX64dRs9

[Aaron Stannard @Aaronontheweb]

New Video / Blog Post: "Run Local Akka .NET Clusters with .NET Aspire"

The biggest win here is that debugging / orchestrating a cluster is trivial now thanks to Aspire + Akka .Management.

Links in next post.

[Akka.NET @AkkaDotNET]

@Aaronontheweb 🚀🚀🚀

[Aaron Stannard @Aaronontheweb]

Codex proposes adding locking, status flags, and a bunch of shared-task coordination inside a hosted service for doing skill sync in Netclaw.

Me: "you know we have access to Akka .NET actors right? Just use those."

* Removed 318 total LOC
* Eliminated several failure classes
* Made the code readable and understandable

You should be using actors more. Your agents should be using them more.