Monitoring, Alerts, and Backup¶

📖 Atlas Monitoring - Monitoring documentation 📖 Atlas Backup - Backup documentation

Monitoring Metrics¶

Cluster Metrics Dashboard¶

📖 Cluster Metrics - Metrics reference

Connection Metrics: | Metric | Description | |--------|-------------| | Connections | Current number of open connections | | Connection Pool Size | Total connections in the pool | | Created Connections | New connections created per second |

Operation Metrics: | Metric | Description | |--------|-------------| | Opcounters | Insert, query, update, delete, getmore, command rates | | Document Metrics | Documents inserted, returned, updated, deleted | | Operation Execution Time | Average execution time per operation type |

Storage Metrics: | Metric | Description | |--------|-------------| | Data Size | Uncompressed logical data size | | Storage Size | On-disk storage including compression | | Index Size | Total index storage size | | Disk IOPS | Read and write I/O operations per second | | Disk Latency | Average disk operation latency | | Disk Space Used | Percentage of provisioned disk used |

Memory and Cache Metrics: | Metric | Description | |--------|-------------| | WiredTiger Cache Used | Percentage of cache in use | | WiredTiger Dirty Cache | Percentage of dirty data in cache | | System Memory | Total and available system memory | | Page Faults | Number of page faults per second |

Replication Metrics: | Metric | Description | |--------|-------------| | Replication Lag | Time behind primary (seconds) | | Replication Headroom | Time before oplog window closes | | Oplog Rate | Rate of oplog entries per hour | | Oplog Window | Time range covered by the oplog |

Network Metrics: | Metric | Description | |--------|-------------| | Bytes In | Network bytes received per second | | Bytes Out | Network bytes sent per second | | Num Requests | Number of requests per second |

Viewing Metrics¶

• Atlas UI: Cluster > Metrics tab
• Real-Time Performance Panel: Live view of active operations
• Atlas CLI: atlas metrics processes <hostname>:<port>
• Atlas Admin API: GET /groups/{groupId}/processes/{processId}/measurements

Time Granularity¶

Alerts¶

Alert Configuration¶

📖 Alert Configuration - Alert setup guide

Alert Components: 1. Target - What to monitor (host, replica set, cluster) 2. Condition - Threshold or event trigger 3. Notification - How to notify (email, Slack, PagerDuty, webhook)

Common Alert Conditions¶

Host Alerts: | Condition | Recommended Threshold | |-----------|----------------------| | CPU % above | 80% for 10 minutes | | Memory % above | 90% for 10 minutes | | Disk % above | 80% | | Connections % above | 80% of max | | Page faults above | Baseline dependent | | Query targeting above | 1000 (docs scanned/returned) |

Replica Set Alerts: | Condition | Description | |-----------|-------------| | Replication lag above | Seconds behind primary | | No primary detected | Primary election failed | | Too few healthy members | Below quorum |

Cluster Alerts: | Condition | Description | |-----------|-------------| | Cluster deleted | Cluster was terminated | | Cluster scaled | Auto-scaling event occurred |

Billing Alerts: | Condition | Description | |-----------|-------------| | Monthly bill estimate above | Spending threshold | | Credit card expiring | Payment method expiring |

Notification Channels¶

Third-Party Integrations¶

📖 Integrations - Integration guide

Cloud Backup¶

Overview¶

📖 Cloud Backup - Backup overview

• Available for M10+ dedicated clusters
• Continuous backup using oplog for point-in-time recovery
• Automated snapshots on configurable schedule
• Stored in the same cloud provider and region as the cluster
• Encrypted at rest

Snapshot Schedule¶

Default Schedule: | Frequency | Retention | |-----------|-----------| | Every 6 hours | 2 days | | Daily | 7 days | | Weekly (Saturday) | 4 weeks | | Monthly (last day) | 12 months |

Customizable Settings: - Snapshot frequency (1, 2, 4, 6, 8, 12, or 24 hours) - Retention period for each frequency - Reference time zone - Reference hour for daily snapshots

Point-in-Time Restore¶

📖 PIT Restore - PIT restore guide

How It Works: 1. Atlas continuously captures oplog entries 2. Oplog window defines how far back you can restore 3. Restore to any second within the oplog window 4. Typically 24-hour oplog window

Restore Options: | Option | Description | |--------|-------------| | Restore to same cluster | Overwrites current data | | Restore to new cluster | Creates a new cluster with restored data | | Download snapshot | Download BSON files for local restore |

Restore Procedure: 1. Navigate to Backup > Continuous Backup 2. Choose "Restore" and select restore type 3. For PIT: Select the exact date and time 4. For snapshot: Select a specific snapshot 5. Choose target (same cluster, new cluster, or download) 6. Confirm and start restore

On-Demand Snapshots¶

• Trigger a snapshot at any time
• Useful before major operations (migrations, deployments)
• Counts toward snapshot retention
• Available via Atlas UI, CLI, or API

# Create on-demand snapshot via CLI
atlas backups snapshots create <clusterName> \
  --desc "Pre-deployment backup"

Backup Compliance Policy¶

📖 Backup Compliance - Compliance settings

Features: - Prevent unauthorized backup deletion - Enforce minimum retention periods - Require point-in-time restore capability - Prevent cluster deletion without backup

Key Points: - Once enabled, cannot be disabled without support - Applies to all clusters in the project - Overrides individual cluster backup policies - Requires Organization Owner to enable

Backup for Sharded Clusters¶

• Snapshots capture all shards at the same point
• Atlas coordinates snapshot timing across shards
• Restore restores the entire sharded cluster
• Cannot restore individual shards

Atlas CLI Monitoring Commands¶

# List processes
atlas processes list --projectId <projectId>

# Get process metrics
atlas metrics processes <hostname>:<port> \
  --granularity PT1M \
  --period PT24H \
  --type CONNECTIONS

# List alerts
atlas alerts list --projectId <projectId>

# Acknowledge alert
atlas alerts acknowledge <alertId>

# List backups
atlas backups snapshots list <clusterName>

# Restore from snapshot
atlas backups restores start <clusterName> \
  --snapshotId <snapshotId> \
  --targetClusterName <targetCluster>

Monitoring Best Practices¶

1. Configure alerts early - Set up before production traffic
2. Start with recommended alerts - CPU, memory, disk, connections, replication lag
3. Set appropriate thresholds - Avoid alert fatigue from overly sensitive alerts
4. Use multiple notification channels - Email + Slack/PagerDuty for redundancy
5. Monitor query targeting - High ratios indicate missing indexes
6. Track metrics over time - Look for trends, not just current values
7. Test restore procedures - Regularly verify backup integrity
8. Configure backup compliance - For production workloads with compliance needs
9. Use on-demand snapshots - Before major changes
10. Review and update alerts - As workload patterns change