Improve Pulse execution reliability: retry logic, better logging, SSH hardening

monitor.py / diagnose.py PulseClient.run_command:
- Add automatic single retry on submit failure, explicit Pulse failure
  (status=failed/timed_out), and poll timeout — handles transient SSH
  or Pulse hiccups without dropping the whole collection cycle
- Log execution_id and full Pulse URL on every failure so failed runs
  can be found in the Pulse UI immediately
- Handle 'timed_out' and 'cancelled' Pulse statuses explicitly (previously
  only 'failed' was caught; others would spin until local deadline)
- Poll every 2s instead of 1s to reduce Pulse API chatter

SSH command options (_ssh_batch + diagnose.py):
- Add BatchMode=yes: aborts immediately instead of hanging on a
  password prompt if key auth fails
- Add ServerAliveInterval=10 ServerAliveCountMax=2: SSH detects a
  hung remote command within ~20s instead of sitting silent until the
  45s Pulse timeout expires

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

diagnose.py

@@ -77,7 +77,9 @@ class DiagnosticsRunner:
 
         return (
             f'ssh -o StrictHostKeyChecking=no -o ConnectTimeout=5 '
-            f'-o LogLevel=ERROR root@{ip_q} \'{remote_cmd}\''
+            f'-o BatchMode=yes -o LogLevel=ERROR '
+            f'-o ServerAliveInterval=10 -o ServerAliveCountMax=2 '
+            f'root@{ip_q} \'{remote_cmd}\''
         )
 
     # ------------------------------------------------------------------

monitor.py

@@ -246,11 +246,16 @@ class PulseClient:
             'Content-Type': 'application/json',
         })
 
-    def run_command(self, command: str) -> Optional[str]:
-        """Submit *command* to Pulse, poll until done, return stdout or None."""
+    def run_command(self, command: str, _retry: bool = True) -> Optional[str]:
+        """Submit *command* to Pulse, poll until done, return stdout or None.
+
+        Retries once automatically on transient submit failures or timeouts.
+        """
         self.last_execution_id = None
         if not self.url or not self.api_key or not self.worker_id:
             return None
+
+        # Submit
         try:
             resp = self.session.post(
                 f'{self.url}/api/internal/command',
@@ -262,11 +267,16 @@ class PulseClient:
             self.last_execution_id = execution_id
         except Exception as e:
             logger.error(f'Pulse command submit failed: {e}')
+            if _retry:
+                logger.info('Retrying Pulse command submit in 5s...')
+                time.sleep(5)
+                return self.run_command(command, _retry=False)
             return None
 
+        # Poll
         deadline = time.time() + self.timeout
         while time.time() < deadline:
-            time.sleep(1)
+            time.sleep(2)
             try:
                 r = self.session.get(
                     f'{self.url}/api/internal/executions/{execution_id}',
@@ -281,11 +291,28 @@ class PulseClient:
                         if entry.get('action') == 'command_result':
                             return entry.get('stdout', '')
                     return ''
-                if status == 'failed':
+                if status in ('failed', 'timed_out', 'cancelled'):
+                    logger.error(
+                        f'Pulse execution {execution_id} ended with status={status!r}; '
+                        f'view at {self.url}/executions/{execution_id}'
+                    )
+                    if _retry and status != 'cancelled':
+                        logger.info('Retrying failed Pulse command in 5s...')
+                        time.sleep(5)
+                        return self.run_command(command, _retry=False)
                     return None
             except Exception as e:
-                logger.error(f'Pulse poll failed: {e}')
-        logger.warning(f'Pulse command timed out after {self.timeout}s')
+                logger.error(f'Pulse poll failed for {execution_id}: {e}')
+
+        logger.warning(
+            f'Pulse command timed out after {self.timeout}s '
+            f'(execution_id={execution_id}); '
+            f'view at {self.url}/executions/{execution_id}'
+        )
+        if _retry:
+            logger.info('Retrying timed-out Pulse command in 5s...')
+            time.sleep(5)
+            return self.run_command(command, _retry=False)
         return None
 
 
@@ -336,7 +363,9 @@ class LinkStatsCollector:
 
         ssh_cmd = (
             f'ssh -o StrictHostKeyChecking=no -o ConnectTimeout=5 '
-            f'-o LogLevel=ERROR root@{ip} "{shell_cmd}"'
+            f'-o BatchMode=yes -o LogLevel=ERROR '
+            f'-o ServerAliveInterval=10 -o ServerAliveCountMax=2 '
+            f'root@{ip} "{shell_cmd}"'
         )
         output = self.pulse.run_command(ssh_cmd)
         if output is None: