Examples Checker

Examples Checker Skill

Use this skill to validate Ankaios examples by building, running, and verifying their expected behavior inside the devcontainer.

Trigger behavior

• If the user asks to "check examples" (or equivalent), run all examples by default.
• If the user specifies a particular example name (e.g. rust_sdk_hello), only check that one.
• If the user asks to "check control interface examples", run only the four control interface examples.
• If the user asks to "check SDK examples", run only the SDK examples.

Example discovery

• Discover example subfolders at runtime by listing /workspaces/ankaios/examples/.
• Exclude tools/ folder.

Example categories

Control Interface examples (raw protobuf)

Expected behavior: Each creates a dynamic_nginx workload via the Control Interface, then polls workload states every 5 seconds. The example workload itself should reach Running(Ok).

SDK examples

Prerequisites check

Before running any example, verify:

1. Ankaios binaries are built and available. Check for debug binaries first:
   • Debug: /workspaces/ankaios/target/x86_64-unknown-linux-musl/debug/
   • Release: /workspaces/ankaios/target/x86_64-unknown-linux-musl/release/
   • If neither exists, build with cargo build.
2. No stale Ankaios processes are running: run the cleanup script before starting.
3. Set ANK_BIN_DIR to the directory containing the Ankaios binaries (debug or release).

Existing scripts

The examples/ folder provides two helper scripts that must be used:

• run_example.sh <example_name> — Builds the container image, starts Ankaios (server + agent_A) if not already running, and applies the example manifest. Accepts --ankaios-bin-dir or uses the ANK_BIN_DIR environment variable.
• cleanup.sh — Stops Ankaios server and agent, stops and removes all podman containers and volumes.

Execution process

Per-example execution

For each example to check:

1. Clean state — Run examples/cleanup.sh to ensure no leftover state.

2. Run the example — From the examples/ directory, run in the foreground and wait for it to finish:

   ./run_example.sh <example_name>

   This script handles: building the container image, starting Ankaios (server + agent) in the background, and applying the manifest. The script returns once the build is done and Ankaios is started. If the build fails, record the error and move to the next example.

3. Wait for workloads — Poll with ank get workloads every 5 seconds, up to 60 seconds, until the example workload reaches Running(Ok).

4. Validate behavior — Perform example-specific validation (see below).

5. Collect logs — Run ank logs <workload_name> to capture workload output.

6. Cleanup — Run examples/cleanup.sh after each example.

Example-specific validation

Control Interface examples (rust_control_interface, cpp_control_interface, nodejs_control_interface, python_control_interface)

After the example workload reaches Running(Ok):

• Wait up to 30 seconds for the dynamic_nginx workload to appear in ank get workloads.
• Verify dynamic_nginx reaches Running(Ok).
• Check example workload logs (ank logs <example_name>) for evidence of successful Control Interface communication (e.g., workload state output).
• Mark as passed if dynamic_nginx was created and reached Running state.

python_sdk_hello / rust_sdk_hello

After the example workload reaches Running(Ok):

• Check workload logs for evidence of:
  1. dynamic_nginx workload being created.
  2. dynamic_nginx reaching a Running state.
  3. dynamic_nginx being deleted.
• The example workload may exit after completion — this is expected.
• Mark as passed if logs show the create → wait → delete lifecycle completed.