support: diagnostic debug for flaky test

This patch adds diagnostic data and additional logging to catch the root
cause for issue #1303:

  - Exactly when and why cleanup is called
  - Whether the cd /tmp command succeeds
  - The actual tar exit code (not just 255 from SSH)
  - Whether cleanup happens before tar completes (race condition)
  - The full collection.log showing the sequence of events

Signed-off-by: Joachim Wiberg <troglobit@gmail.com>

src/bin/support

@@ -129,8 +129,12 @@ cmd_collect()
     # Cleanup on exit
     cleanup()
     {
+        echo "[$(date -Iseconds)] Cleanup called (signal: ${1:-EXIT})" >> "${EXEC_LOG}" 2>&1 || echo "[$(date -Iseconds)] Cleanup called (signal: ${1:-EXIT})" >&2
         if [ -d "${COLLECT_DIR}" ]; then
+            echo "[$(date -Iseconds)] Removing collection directory: ${COLLECT_DIR}" >> "${EXEC_LOG}" 2>&1 || echo "[$(date -Iseconds)] Removing: ${COLLECT_DIR}" >&2
             rm -rf "${COLLECT_DIR}"
+        else
+            echo "[$(date -Iseconds)] Collection directory already gone: ${COLLECT_DIR}" >> "${EXEC_LOG}" 2>&1 || echo "[$(date -Iseconds)] Already gone: ${COLLECT_DIR}" >&2
         fi
     }
     trap cleanup EXIT INT TERM
@@ -383,7 +387,14 @@ cmd_collect()
 
     # Create final tar.gz and output to stdout
     # Use -C to change to parent directory so paths in archive don't include full path
-    cd "${WORK_DIR}"
+    echo "[$(date -Iseconds)] Changing to work directory: ${WORK_DIR}" >> "${EXEC_LOG}" 2>&1
+    if ! cd "${WORK_DIR}"; then
+        echo "[$(date -Iseconds)] ERROR: Failed to cd to ${WORK_DIR}" >> "${EXEC_LOG}" 2>&1
+        echo "Error: Cannot change to work directory ${WORK_DIR}" >&2
+        exit 1
+    fi
+    echo "[$(date -Iseconds)] Successfully changed to: $(pwd)" >> "${EXEC_LOG}" 2>&1
+    echo "[$(date -Iseconds)] Creating archive from: $(basename "${COLLECT_DIR}")" >> "${EXEC_LOG}" 2>&1
 
     # Check if password encryption is requested
     if [ -n "$PASSWORD" ]; then
@@ -392,14 +403,29 @@ cmd_collect()
             exit 1
         fi
         echo "Encrypting with GPG..." >&2
+        echo "[$(date -Iseconds)] Starting tar with GPG encryption" >> "${EXEC_LOG}" 2>&1
         tar czf - "$(basename "${COLLECT_DIR}")" 2>> "${EXEC_LOG}" | \
             gpg --batch --yes --passphrase "$PASSWORD" --pinentry-mode loopback -c 2>> "${EXEC_LOG}"
+        tar_exit=$?
+        echo "[$(date -Iseconds)] tar+gpg pipeline exit code: $tar_exit" >> "${EXEC_LOG}" 2>&1
         echo "" >&2
         echo "WARNING: Remember to share the encryption password out-of-band!" >&2
         echo "         Do not send it in the same email as the encrypted file." >&2
+        if [ $tar_exit -ne 0 ]; then
+            echo "[$(date -Iseconds)] ERROR: tar+gpg failed with exit code $tar_exit" >> "${EXEC_LOG}" 2>&1
+            exit $tar_exit
+        fi
     else
+        echo "[$(date -Iseconds)] Starting tar (no encryption)" >> "${EXEC_LOG}" 2>&1
         tar czf - "$(basename "${COLLECT_DIR}")" 2>> "${EXEC_LOG}"
+        tar_exit=$?
+        echo "[$(date -Iseconds)] tar exit code: $tar_exit" >> "${EXEC_LOG}" 2>&1
+        if [ $tar_exit -ne 0 ]; then
+            echo "[$(date -Iseconds)] ERROR: tar failed with exit code $tar_exit" >> "${EXEC_LOG}" 2>&1
+            exit $tar_exit
+        fi
     fi
+    echo "[$(date -Iseconds)] Archive creation completed successfully" >> "${EXEC_LOG}" 2>&1
 }
 
 cmd_clean()

test/case/misc/support_collect/test.py

@@ -48,6 +48,21 @@ with infamy.Test() as test:
             stderr_output = result.stderr.decode('utf-8') if result.stderr else ""
             print(f"support collect failed with return code {result.returncode}")
             print(f"stderr: {stderr_output}")
+
+            # Try to retrieve the collection.log for debugging
+            print("\n=== Attempting to retrieve collection.log for debugging ===")
+            try:
+                log_result = tgtssh.run("find /tmp -name 'support-*' -type d -exec cat {}/collection.log \\; 2>/dev/null || echo 'No collection.log found'",
+                                       stdout=subprocess.PIPE,
+                                       stderr=subprocess.PIPE,
+                                       timeout=10,
+                                       check=False)
+                if log_result.stdout:
+                    log_output = log_result.stdout.decode('utf-8')
+                    print(f"collection.log contents:\n{log_output}")
+            except Exception as e:
+                print(f"Could not retrieve collection.log: {e}")
+
             raise Exception("support collect command failed")
 
     with test.step("Verify tarball was created and is valid"):
@@ -110,6 +125,21 @@ with infamy.Test() as test:
             if result.returncode != 0:
                 stderr_output = result.stderr.decode('utf-8') if result.stderr else ""
                 print(f"support collect with encryption failed: {stderr_output}")
+
+                # Try to retrieve the collection.log for debugging
+                print("\n=== Attempting to retrieve collection.log for debugging ===")
+                try:
+                    log_result = tgtssh.run("find /tmp -name 'support-*' -type d -exec cat {}/collection.log \\; 2>/dev/null || echo 'No collection.log found'",
+                                           stdout=subprocess.PIPE,
+                                           stderr=subprocess.PIPE,
+                                           timeout=10,
+                                           check=False)
+                    if log_result.stdout:
+                        log_output = log_result.stdout.decode('utf-8')
+                        print(f"collection.log contents:\n{log_output}")
+                except Exception as e:
+                    print(f"Could not retrieve collection.log: {e}")
+
                 raise Exception("support collect with --password failed")
 
         with test.step("Verify encrypted file and decrypt it"):