Is there an existing issue for this?
Describe the bug
In previous releases there was startup time regression in keycloak and was attributed to issues in other areas.
Those issues are reported to have been fixed, and yet we still see 3+ minutes startup time for Keycloak.
This slowdown is hurting our e2e tests in CI. We prepull the image before aspire start so the only cost is the actual container startup time.
I tried to reproduce it with a minimal apphost, and it only took 23 seconds, so probably something about the orchestration of our apphost is making the keycloak startup slow.
In local development I can see that it takes a while before aspire tries to start the docker, and even after it is started it takes a long time.
Expected Behavior
starting keycloak takes 23 seconds in an empty apphost - the startup time in a larger apphost shouldn't be substantially longer (less than 1 minute).
Steps To Reproduce
I created an empty apphost with only keycloak and it started up in 23 seconds.
In our large apphost it took over 3 minutes in CI, and locally 2 minutes.
Exceptions (if any)
No response
Aspire doctor output
➜ aspire doctor
Aspire Environment Check
========================
Aspire
✅ Aspire CLI version 13.5.3 (channel: stable)
✅ Developer Control Plane (DCP) connection health checks succeeded
AppHost
✅ AppHost version 13.5.3 (apphost.cs)
.NET SDK
✅ .NET 10.0.203 installed (arm64)
Container Runtime
✅ Docker v29.1.3: running (auto-detected (default)) ← active
Environment
✅ Operating system: macOS 15.7.7
✅ HTTPS development certificate is trusted
Development Tools
✅ Aspire extension for VS Code is installed
Summary: 8 passed, 0 warnings, 0 failed
Aspire CLI Installations
========================
╭───────────────────────────────────────────┬─────────────────────────────────────────────────┬─────────┬───────────┬─────────────╮
│ Path │ Version │ Channel │ Route │ PATH status │
├───────────────────────────────────────────┼─────────────────────────────────────────────────┼─────────┼───────────┼─────────────┤
│ /Users/twald/.aspire/bin/aspire (current) │ 13.5.3+b5f143315ffb6968ea939a9978797a5b20e4c688 │ stable │ (unknown) │ active │
╰───────────────────────────────────────────┴─────────────────────────────────────────────────┴─────────┴───────────┴─────────────╯
Anything else?
No response
Is there an existing issue for this?
Describe the bug
In previous releases there was startup time regression in keycloak and was attributed to issues in other areas.
Those issues are reported to have been fixed, and yet we still see 3+ minutes startup time for Keycloak.
This slowdown is hurting our e2e tests in CI. We prepull the image before
aspire startso the only cost is the actual container startup time.I tried to reproduce it with a minimal apphost, and it only took 23 seconds, so probably something about the orchestration of our apphost is making the keycloak startup slow.
In local development I can see that it takes a while before aspire tries to start the docker, and even after it is started it takes a long time.
Expected Behavior
starting keycloak takes 23 seconds in an empty apphost - the startup time in a larger apphost shouldn't be substantially longer (less than 1 minute).
Steps To Reproduce
I created an empty apphost with only keycloak and it started up in 23 seconds.
In our large apphost it took over 3 minutes in CI, and locally 2 minutes.
Exceptions (if any)
No response
Aspire doctor output
Anything else?
No response