Fix gather-extra post-phase failures to show contextual message instead of re-trigger suggestion

Author: qu4rkn3t

bugzooka/analysis/failure_keywords.py

@@ -4,6 +4,7 @@
     "provision": ("Provision", "provision failure"),
     "deprovision": ("Deprovision", "deprovision failure"),
     "gather": ("Must Gather", "must gather failure"),
+    "gather-extra": ("Gather Extra", "gather-extra failure"),
     "orion": ("Changepoint", "orion changepoint detection failure"),
     "cerberus": ("Prow Other", "cerberus health check failure"),
     "node-readiness": ("Node Readiness", "node-readiness check failure"),

bugzooka/analysis/prow_analyzer.py

@@ -292,13 +292,15 @@ def analyze_prow_artifacts(directory_path, job_name):
             step_summary = ""
     if matched_line is None:
         if step_name and step_phase:
+            # Determine if it's an install issue based on the phase
+            # Post-phase runs after the main workload, so failures there aren't install issues
+            is_install_issue = step_phase.lower() not in ["post", "test"]
+
             return ProwAnalysisResult(
                 errors=[step_summary] if step_summary else [],
-                categorization_message=(
-                    f"Maintenance Issue: {step_name} failed in" f" {step_phase} phase"
-                ),
+                categorization_message=categorization_message,  # Use already-set categorization
                 requires_llm=False,
-                is_install_issue=True,
+                is_install_issue=is_install_issue,
                 step_name=step_name,
                 full_errors_for_file=None,
             )

bugzooka/integrations/slack_fetcher.py

@@ -260,7 +260,27 @@ def _send_error_logs_preview(
         # in the thread (just before job history).
         pending_file = errors_for_file.strip() if needs_file else None
 
-        if is_install_issue:
+        # Check for post-phase and gather-extra failures
+        is_post_phase = categorization_message and "post phase" in categorization_message.lower()
+        is_gather_extra = categorization_message and "gather-extra" in categorization_message.lower()
+
+        if is_post_phase and is_gather_extra:
+            # gather-extra runs after the main workload, so provide context
+            retrigger_message = (
+                "gather-extra runs at the end of the pipeline. "
+                "Appears to be general workload failure, but we have performance results to look further."
+            )
+            message_block = self.get_slack_message_blocks(
+                markdown_header=":information_source: *Post-Phase Context*\n",
+                content_text=retrigger_message,
+            )
+            self.client.chat_postMessage(
+                channel=self.channel_id,
+                text="Post-Phase Context",
+                blocks=message_block,
+                thread_ts=max_ts,
+            )
+        elif is_install_issue:
             retrigger_message = (
                 "This appears to be an installation or maintenance issue. "
                 "Please re-trigger the run."