Proceed with backport/rebase when build times out

This is a short term solution, in future we might rather do async builds/skip the builds completely and iterate based on CI.

Author: lbarcziova

agents/backport_agent.py

@@ -313,6 +313,9 @@ async def run_build_agent(state):
                 build_result = BuildOutputSchema.model_validate_json(response.last_message.text)
                 if build_result.success:
                     return "update_release"
+                if build_result.is_timeout:
+                    logger.info(f"Build timed out for {state.jira_issue}, proceeding")
+                    return "update_release"
                 state.attempts_remaining -= 1
                 if state.attempts_remaining <= 0:
                     state.backport_result.success = False

agents/build_agent.py

@@ -34,8 +34,7 @@ def get_instructions() -> str:
       using the `download_artifacts` tool to the current directory. If there are no log files,
       just return the error message. Otherwise, start with `builder-live.log` and try to identify
       the build failure. If not found, try the same with `root.log`. Summarize the findings
-      and return them as `error`.
-
+      and return them as `error`. If the build failed due to a timeout, set `is_timeout` to `true` in your output.
 
       General instructions:
 

agents/rebase_agent.py

@@ -274,6 +274,9 @@ async def run_build_agent(state):
                 build_result = BuildOutputSchema.model_validate_json(response.last_message.text)
                 if build_result.success:
                     return "update_release"
+                if build_result.is_timeout:
+                    logger.info(f"Build timed out for {state.jira_issue}, proceeding")
+                    return "update_release"
                 state.attempts_remaining -= 1
                 if state.attempts_remaining <= 0:
                     state.rebase_result.success = False

common/models.py

@@ -242,6 +242,7 @@ class BuildOutputSchema(BaseModel):
     """Output schema for the build agent."""
     success: bool = Field(description="Whether the build was successfully completed")
     error: str | None = Field(description="Specific details about an error")
+    is_timeout: bool = Field(default=False, description="Whether the build failed due to a timeout")
 
 
 # ============================================================================