orleans | Cloud Native application framework for NET | Architecture library

In base Python, we can try sorting using a lambda expression:

Thanks to @BozoJoe's comment I could debug this.

The problem was that it was trying to scrape ports 30000 and 1111 instead of 8082 like I said before. I could see this thanks to the Prometheus dashboard at localhost:9090/targets

So I went to prometheus config file and make sure to start scrapping the correct port (also I added some restrictions to the search for name):

Introduction of a complex return type to minimize roundtrips is a valid solution.

However, I wouldn't return the whole internal state, as I assume that not all the clients need all the data all the time. It may also be a sign that you have a business logic implemented outside of the grains and you should move it into the grain.

With map_data, you need to use county rather than state to get the correct subregions. Then, we can use left_join to merge them together by county (i.e., subregion and Jurisdiction). There is a letter case difference, so I first converted Jurisdiction to lowercase to match the data from map_data. Note: df is the OP data from dput.

To align the columns, set font of each button to "Courier" (or another monospaced font). There are at least two ways to do this:

1. when creating the button
2. assign the button to a variable and then call configure on it.

Cloud Run is a serverless product. As serverless product, you don't manage the servers and the network. Therefore, asking for the podIP make no sense and has no value (at least, it will be a private IP in the Google Cloud serverless network).

So, the answer is "NO, you can't get the podIP".

But why do you need it? You can set a dummy value and see what happens we your container. Or to update your code to make optional that value.

Your understanding is close but missing a key aspect of how Orleans finds and places grain activations.

The key is that most placement providers in Orleans rely on a grain directory. That directory stores the current mapping of which grain is activated on which server. Like placement, the directory is configurable. It uses an in-cluster directory by default, where responsibility for storing the mapping is shared among the servers.

Orleans will check the directory first, to see if the grain is currently activated somewhere in the cluster and it will only run the logic in PreferLocalPlacement when the grain does not currently have a valid entry in the directory.

Therefore, you will have one instance of your grain active in the cluster, not one instance per server. If there are no currently active instances of your grain then PreferLocalPlacement will always pick the current, local server (or the server which the request first lands on) as long as it is compatible.

Clusters can have a mix of servers each having a different set of grain assemblies loaded. Therefore, placement providers need to take into account whether a given server is compatible with the requested grain type and if it has a compatible version of the interface which is being requested. If it does not support that grain, then a random compatible server will be chosen instead. For more information, see the documentation on Heterogeneous Clusters.

You might ask which placement providers do not rely on a grain directory. The only answer today is [StatelessWorker] placement, since [StatelessWorker] grains can have multiple instances, i.e, each server can have a configurable number of local instances of any given [StatelessWorker] grain.

The setup needs to be improved. You can check this sample from the official Orleans GitHub page: https://github.com/dotnet/orleans/blob/main/test/TesterInternal/StreamingTests/SMSStreamingTests.cs

So, in your ClusterFixture class you should add the SiloConfigurator:

It was a kind of user error : I was using the same deployment id for my local development and Azure deployment, and they each other could not confirm dead or alive, messing up the service.

I have separated the cluster id for my local development and Azure deployment, as in the screenshot :

Membership Table

Then everything works fine. Although still Azure function app restarts itself occasionally, they detect the dead silo immediately and clean up the states.