[UX]: Make run status in UI and CLI easier to understand #2716

Author: peterschmidt85

src/dstack/_internal/core/models/runs.py

@@ -351,30 +351,20 @@ def _error(cls, values) -> Dict:
 
     @staticmethod
     def _get_error(termination_reason: Optional[JobTerminationReason]) -> Optional[str]:
-        if termination_reason == JobTerminationReason.INSTANCE_UNREACHABLE:
-            return "instance unreachable"
-        elif termination_reason == JobTerminationReason.WAITING_INSTANCE_LIMIT_EXCEEDED:
-            return "waiting instance limit exceeded"
-        elif termination_reason == JobTerminationReason.VOLUME_ERROR:
-            return "waiting runner limit exceeded"
-        elif termination_reason == JobTerminationReason.GATEWAY_ERROR:
-            return "gateway error"
-        elif termination_reason == JobTerminationReason.SCALED_DOWN:
-            return "scaled down"
-        elif termination_reason == JobTerminationReason.INACTIVITY_DURATION_EXCEEDED:
-            return "inactivity duration exceeded"
-        elif termination_reason == JobTerminationReason.TERMINATED_DUE_TO_UTILIZATION_POLICY:
-            return "utilization policy"
-        elif termination_reason == JobTerminationReason.PORTS_BINDING_FAILED:
-            return "ports binding failed"
-        elif termination_reason == JobTerminationReason.CREATING_CONTAINER_ERROR:
-            return "runner error"
-        elif termination_reason == JobTerminationReason.EXECUTOR_ERROR:
-            return "executor error"
-        elif termination_reason == JobTerminationReason.MAX_DURATION_EXCEEDED:
-            return "max duration exceeded"
-        else:
-            return None
+        error_mapping = {
+            JobTerminationReason.INSTANCE_UNREACHABLE: "instance unreachable",
+            JobTerminationReason.WAITING_INSTANCE_LIMIT_EXCEEDED: "waiting instance limit exceeded",
+            JobTerminationReason.VOLUME_ERROR: "waiting runner limit exceeded",
+            JobTerminationReason.GATEWAY_ERROR: "gateway error",
+            JobTerminationReason.SCALED_DOWN: "scaled down",
+            JobTerminationReason.INACTIVITY_DURATION_EXCEEDED: "inactivity duration exceeded",
+            JobTerminationReason.TERMINATED_DUE_TO_UTILIZATION_POLICY: "utilization policy",
+            JobTerminationReason.PORTS_BINDING_FAILED: "ports binding failed",
+            JobTerminationReason.CREATING_CONTAINER_ERROR: "runner error",
+            JobTerminationReason.EXECUTOR_ERROR: "executor error",
+            JobTerminationReason.MAX_DURATION_EXCEEDED: "max duration exceeded",
+        }
+        return error_mapping.get(termination_reason)
 
 
 class Job(CoreModel):

src/tests/_internal/core/models/test_runs.py

@@ -1,6 +1,8 @@
 from dstack._internal.core.models.runs import (
     JobStatus,
+    JobSubmission,
     JobTerminationReason,
+    Run,
     RunStatus,
     RunTerminationReason,
 )
@@ -22,3 +24,37 @@ def test_job_termination_reason_to_status_works_with_all_enum_varians():
     for job_termination_reason in JobTerminationReason:
         job_status = job_termination_reason.to_status()
         assert isinstance(job_status, JobStatus)
+
+
+# Will fail if JobTerminationReason value is added without updaing JobSubmission._get_error
+def test_get_error_returns_expected_messages():
+    no_error_reasons = [
+        JobTerminationReason.FAILED_TO_START_DUE_TO_NO_CAPACITY,
+        JobTerminationReason.INTERRUPTED_BY_NO_CAPACITY,
+        JobTerminationReason.WAITING_RUNNER_LIMIT_EXCEEDED,
+        JobTerminationReason.TERMINATED_BY_USER,
+        JobTerminationReason.DONE_BY_RUNNER,
+        JobTerminationReason.ABORTED_BY_USER,
+        JobTerminationReason.TERMINATED_BY_SERVER,
+        JobTerminationReason.CONTAINER_EXITED_WITH_ERROR,
+    ]
+
+    for reason in JobTerminationReason:
+        if JobSubmission._get_error(reason) is None:
+            # Fail no-error reason is not in the list
+            assert reason in no_error_reasons
+
+
+# Will fail if RunTerminationReason value is added without updating Run._get_error
+def test_run_get_error_returns_none_for_specific_reasons():
+    no_error_reasons = [
+        RunTerminationReason.ALL_JOBS_DONE,
+        RunTerminationReason.JOB_FAILED,
+        RunTerminationReason.STOPPED_BY_USER,
+        RunTerminationReason.ABORTED_BY_USER,
+    ]
+
+    for reason in RunTerminationReason:
+        if Run._get_error(reason) is None:
+            # Fail no-error reason is not in the list
+            assert reason in no_error_reasons