Fix AttributeError in cluster metrics recording when connection is None or ClusterNode object instance is used to extract the connection info

[petyaslavova]

Fix AttributeError in cluster metrics recording when connection is None

When exceptions occur in RedisCluster before a connection is established
(e.g., AuthenticationError, MaxConnectionsError), the connection variable
is None or a ClusterNode object (which lacks a 'db' attribute). This caused
AttributeError when recording metrics.

Changes:

• Use e.connection (which is set to target_node) instead of connection
  variable for metrics recording in exception handlers
• Add defensive checks in _record_command_metric for missing attributes
• ClusterNode objects now default to db="0" for metrics

Fixes #3997

---

Note

Medium Risk
Touches Redis Cluster command execution error paths and observability hooks; functional behavior should be unchanged but regressions could affect error handling/metrics accuracy during connection establishment failures.

Overview
Prevents AttributeError during Redis Cluster observability by consistently recording command duration metrics using e.connection (which may be a ClusterNode fallback) across exception handlers, including AuthenticationError, MaxConnectionsError, and connection/timeout-related failures.

Makes _record_command_metric defensive when connection is None or lacks db, defaulting host/port/db to safe values, and expands tests to cover sync and asyncio cluster scenarios for these error cases and successful metric recording with a non-default DB.

^{Written by Cursor Bugbot for commit bc9ddb4. This will update automatically on new commits. Configure here.}

[jit-ci]

🛡️ Jit Security Scan Results

✅ No security findings were detected in this PR

---

^{Security scan by Jit}