From 5816fe2eb3f55e1b30ab4906e75bb7870afcdaa3 Mon Sep 17 00:00:00 2001
From: Jeffrey Phillips Freeman <the@jeffreyfreeman.me>
Date: Mon, 23 Feb 2026 16:38:17 +0000
Subject: [PATCH] fix(test): harden database_integration robot suite against CI
 flakiness

- Replace NamedTemporaryFile with mkstemp to avoid leaking an open fd
- Increase subprocess timeout from 30s to 60s for CI headroom
- Retry once on empty-output failure (signal-killed by OOM/CPU starvation)
- Include return code in failure message for easier diagnosis
---
 robot/database_integration.robot | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)

diff --git a/robot/database_integration.robot b/robot/database_integration.robot
index 0c07dac5a..67e25f914 100644
--- a/robot/database_integration.robot
+++ b/robot/database_integration.robot
@@ -716,7 +716,10 @@ Apply Changes With Service
     RETURN    ${count}
 
 Run Python Script
-    [Documentation]    Run a Python script and return output
+    [Documentation]    Run a Python script and return output.
+    ...    Retries once on transient failure (empty output with non-zero exit)
+    ...    to handle CI resource contention when many parallel workers spawn
+    ...    Python subprocesses simultaneously.
     [Arguments]    @{lines}
     # Join lines
     ${script}=    Catenate    SEPARATOR=\n    @{lines}
@@ -752,15 +755,26 @@ Run Python Script
     ${header}=    Fix Python Indentation    ${header_raw}
     # Combine header and code
     ${full_code}=    Catenate    SEPARATOR=\n    ${header}    ${code}
-    # Write to a temporary file
-    ${temp_file}=    Evaluate    __import__('tempfile').NamedTemporaryFile(mode='w', suffix='.py', delete=False, dir='/tmp').name
+    # Write to a temporary file — use mkstemp so the fd is closed before
+    # Create File writes to it (avoids leaking an open descriptor).
+    ${temp_file}=    Evaluate    (lambda t: (__import__('os').close(t[0]), t[1])[-1])(__import__('tempfile').mkstemp(suffix='.py', dir='/tmp'))
     Create File    ${temp_file}    ${full_code}
-    ${result}=    Run Process    ${PYTHON}    ${temp_file}    timeout=30s    stderr=STDOUT    env:PYTHONWARNINGS=ignore    env:PYTHONDONTWRITEBYTECODE=1
+    ${result}=    Run Process    ${PYTHON}    ${temp_file}    timeout=60s    stderr=STDOUT    env:PYTHONWARNINGS=ignore    env:PYTHONDONTWRITEBYTECODE=1
+    # Retry once when the process was likely killed by a signal (OOM / CPU
+    # starvation under heavy parallel CI load).  A signal-killed process
+    # produces empty stdout because its output buffer is never flushed.
+    IF    ${result.rc} != 0
+        ${stdout_len}=    Get Length    ${result.stdout}
+        IF    ${stdout_len} == 0
+            Sleep    3s    reason=Retrying after transient subprocess failure (rc=${result.rc})
+            ${result}=    Run Process    ${PYTHON}    ${temp_file}    timeout=60s    stderr=STDOUT    env:PYTHONWARNINGS=ignore    env:PYTHONDONTWRITEBYTECODE=1
+        END
+    END
     Remove File    ${temp_file}
     # Check if process failed and log stderr if present
     IF    ${result.rc} != 0
-        Log    Process failed with stdout: ${result.stdout}    WARN
-        Fail    Python script execution failed: ${result.stdout}
+        Log    Process failed with rc=${result.rc} stdout: ${result.stdout}    WARN
+        Fail    Python script execution failed (rc=${result.rc}): ${result.stdout}
     END
     # Extract just the number from the output (last line)
     ${lines}=    Split String    ${result.stdout}    \n
-- 
2.52.0