-
Notifications
You must be signed in to change notification settings - Fork 526
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add support for managed identity in Azure Cosmos DB hosting component #7092
Conversation
@@ -205,7 +156,7 @@ public static IResourceBuilder<AzureCosmosDBEmulatorResource> WithGatewayPort(th | |||
/// <param name="builder">Builder for the Cosmos emulator container</param> | |||
/// <param name="count">Desired partition count.</param> | |||
/// <returns>Cosmos emulator resource builder.</returns> | |||
/// <remarks>Not calling this method will result in the default of 25 partitions. The actual started partitions is always one more than specified. | |||
/// <remarks>Not calling this method will result in the default of 10 partitions. The actual started partitions is always one more than specified. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
11 (actual) is the default number of partitions when nothing is set, so updated to 10 since the actual is argument + 1
src/Components/Aspire.Microsoft.Azure.Cosmos/MicrosoftAzureCosmosSettings.cs
Outdated
Show resolved
Hide resolved
namespace Aspire.Hosting.Azure.CosmosDB; | ||
|
||
/// <summary> | ||
/// This health check also creates default databases and containers for the Azure CosmosDB Emulator. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This feels like too much responsibility. Everytime we are checking for "health" we are rechecking all the dbs and containers.
I think it would be better if the health check was only a health check. And then we "waited" for it to be healthy and only ensure the db and containers are created once.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The default health check is checking only one db and its containers (because its state is only for one db, I assume one could create more instances for other dbs). At least we could do the same, pick the first db.
But I did what you suggested at first, though this could introduce undesired effects that it's stated as healthy but we haven't created the db/containers yet. So while we are creating them other services could have started using them.
Another option would be to change the implementation to do like it does right now, but then store a bool and only check for the first db afterwards (not even the containers).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Added a flag to only check/create for the resources once.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This seems like it's the only point of contention in the PR. Is it resolved? The other changes look good. The health check creating the database feels so wrong though 😄 .
This is something we did not solve in the Aspire 9.0 timeframe that we should revisit now.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We should have a live chat about it, I tried different options. Right now it does the creation attempt only once then falls back to the standard check. This is just the apphost health check though to report for WaithHealthyAsync
, the clients use the standard health check.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I approved the PR, can you open an issue? It's kinda related to #7101
src/Aspire.Hosting.Azure.CosmosDB/AzureCosmosDBEmulatorHealthCheck.cs
Outdated
Show resolved
Hide resolved
tests/Aspire.Hosting.Azure.Tests/AzureCosmosDBEmulatorFunctionalTests.cs
Outdated
Show resolved
Hide resolved
tests/Aspire.Hosting.Azure.Tests/AzureCosmosDBEmulatorFunctionalTests.cs
Outdated
Show resolved
Hide resolved
tests/Aspire.Hosting.Azure.Tests/AzureCosmosDBEmulatorFunctionalTests.cs
Outdated
Show resolved
Hide resolved
/azd run |
/azd run |
/azp run |
Azure Pipelines successfully started running 1 pipeline(s). |
Do we want this to be a breaking change? |
At some point, yes. We want managed identity to be the default everywhere, and this is the last integration that still uses secrets by default. So the question in my mind is: "When do we want to make the breaking change?". I think now is a good time. But I can see an argument to wait until the next major version. |
# Conflicts: # playground/CosmosEndToEnd/CosmosEndToEnd.ApiService/Program.cs # playground/CosmosEndToEnd/CosmosEndToEnd.AppHost/Program.cs # src/Aspire.Hosting.Azure.CosmosDB/AzureCosmosDBExtensions.cs # src/Aspire.Hosting.Azure.CosmosDB/AzureCosmosDBResource.cs # src/Aspire.Hosting.Azure.CosmosDB/PublicAPI.Unshipped.txt # tests/Aspire.Hosting.Azure.Tests/AzureCosmosDBEmulatorFunctionalTests.cs
Can we decide if we are going to make the breaking change in 9.1? If yes, we should make sure it's easy to get back to the previous state (and we should validate it) |
This is done with Here is the test that verifies that adding it behaves the same way as before in terms of bicep resources. I also validated it in playground. |
OK let's go with it. Open up a breaking change issue? |
/azp run |
Azure Pipelines successfully started running 1 pipeline(s). |
a78e929
to
518efec
Compare
Description
Fixes #3792
Checklist
<remarks />
and<code />
elements on your triple slash comments?breaking-change
template):doc-idea
template):