Monitor In-Memory DB v2 Clusters Using Grafana

Overview

This tutorial demonstrates how to configure observability for In-Memory DB v2 clusters on IONOS CLOUD using the Logging Service and Monitoring Service. Once both services are activated at the contract level and the cluster level, metrics and logs stream to a regional Grafana endpoint where you can search and visualize In-Memory DB health metrics.

In-Memory DB v2 on IONOS CLOUD provides a set of observable metrics that describe the following:

• Infrastructure health, including CPU time across idle, user, system, and I/O wait states, system load averages, memory availability, disk throughput, IOPS per device, I/O saturation ratios, and filesystem capacity per mount point.
• Database engine health, including instance reachability and uptime, memory usage and fragmentation, client connections, command throughput and latency, cache hit and miss counts, key statistics, persistence state, and command errors.

In-Memory DB v2 metrics provide actionable insight into cluster performance, enabling reliable operations, faster incident resolution, and data-driven capacity planning for production deployments. For the complete list of available metrics, see Metrics.

Target audience

This tutorial targets database administrators, DevOps engineers, and platform engineers who manage In-Memory DB workloads on IONOS CLOUD. Readers benefit from prior experience with REST APIs, Bearer token authentication, and basic PromQL or Grafana dashboard concepts.

What you will learn

By following this tutorial, you will learn how to:

• Activate observability services and allow log and metric collection on an In-Memory DB v2 cluster.
• Access the regional Grafana endpoint and search for In-Memory DB metrics.
• Apply recommended alert thresholds for CPU, memory, disk, and cache metrics.

Prerequisites

Ensure you have:

• An active IONOS CLOUD contract with permissions to manage observability and database resources.
• A valid IONOS CLOUD API token exported as IONOS_TOKEN.
• An In-Memory DB v2 cluster, or permission to create one. For more information, see In-Memory DB v2 API specification.
• curl or any HTTP client for API calls.
• A web browser to use Grafana.

Cost considerations

The following resources are billable and incur costs when you use them:

• Logging and Monitoring services: Charges apply per region after activation.
• In-Memory DB v2 clusters: Compute and snapshot storage charges apply independent of observability settings.

Note: Delete clusters and make observability services inactive if not in use to avoid ongoing charges.

Procedure

The activation flow has two independent layers. You activate the Logging and Monitoring services once per region at the contract level, and you toggle log and metric collection per cluster. Both layers must show true before data appears in Grafana.

Use the following procedure to configure observability for an In-Memory DB v2 cluster:

1

Activate the Logging Service for a region

Send a request to the regional Logging API endpoint with enabled set to true. The Logging Service activates for the contract in that region.

Refer to the full schema in the Logging API v1 documentation.

Note: You can also allow Central Logging through the DCD. For more information, see Send Logs to the Platform.

curl --request PUT \
  --url 'https://logging.de-fra.ionos.com/central' \
  --header 'Authorization: Bearer $IONOS_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{
    "properties": {
      "enabled": true
    }
  }'

The response returns a resource that includes the grafanaEndpoint URL and the enabled: true property. The grafanaEndpoint value is identical for all customers in that region.

2

Activate the Monitoring Service for the same region

Repeat the activation against the Monitoring API endpoint. The Monitoring Service activates separately from Logging.

Refer to the full schema in the Monitoring API v1 documentation.

curl --request PUT \
  --url 'https://monitoring.de-fra.ionos.com/central' \
  --header 'Authorization: Bearer $IONOS_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{
    "properties": {
      "enabled": true
    }
  }'

Note:

• Activation is regional. You can activate Logging and Monitoring in every region where In-Memory DB clusters run. The grafanaEndpoint returned by both services resolves to the same Grafana instance per region, for example https://grafana.logging.de-fra.ionos.com.
• You can perform activation using the DCD. To do so, refer to the full schema in the Define logging pipeline properties and Send Metrics to the Platform.

3

Allow logs and metrics on an In-Memory DB cluster

Set logsEnabled and metricsEnabled to true in the cluster properties block. Apply the setting in one of two ways:

• POST /clusters to create a new cluster with observability activated.
• PUT /clusters/{clusterId} to update an existing cluster. PUT replaces the entire resource. Include all required fields in the request body, even if you are not changing their values. Any field omitted is set to its schema default or cleared to empty.

No separate activation endpoint exists for the cluster-level toggles.

{
  "properties": {
    "name": "my-cluster",
    "version": "9.0",
    "instances": {
      "count": 1,
      "cores": 4,
      "ram": 8
    },
    "connection": {
      "datacenterId": "5a029f4a-72e5-11ec-90d6-0242ac120003",
      "lanId": "2",
      "primaryInstanceAddress": "192.168.2.101/24"
    },
    "persistenceMode": "RDB",
    "evictionPolicy": "allkeys-lru",
    "snapshot": {
      "location": "eu-central-3",
      "retentionDays": 14,
      "snapshotHours": [0, 6, 12, 18]
    },
    "maintenanceWindow": {
      "dayOfTheWeek": "Sunday",
      "time": "02:00:00"
    },
    "logsEnabled": true,
    "metricsEnabled": true
  }
}

Each cluster manages logs and metrics independently. Use it to keep test clusters quiet while production clusters stream full telemetry.

Warning:

• If you set logsEnabled or metricsEnabled to true while the contract-level Logging or Monitoring Service is inactive, the API accepts the request without error.
• No data flows to Grafana until you activate the corresponding service in that region.

4

Verify the activation matrix

Both layers must report true for data to appear in Grafana. The following matrix summarizes every combination.

Logging or Monitoring Service activated	In-Memory DB cluster logs or metrics activated	Visibility result
true	true	Logs visible, metrics visible
true	false	Not visible
false	true	Not visible
false	false	Not visible

5

Access the Grafana dashboard

1. Open the regional Grafana URL returned by the activation response. The URL follows the pattern https://grafana.logging.{region}.ionos.com, for example:

https://grafana.logging.de-fra.ionos.com

Important: Metric collection starts when you provision a cluster. The first metrics appear in Grafana or through the API within 10 to 60 minutes.

2. Sign in with your IONOS CLOUD contract user credentials. Administrator users and users with observability access can sign in.

Note: If you have not previously used the observability product, you cannot sign in to Grafana until your permissions propagate.

3. On the left-hand menu, go to Connections > Data sources, then select Drilldown > Metrics and choose the Mimir data source.

Grafana metrics Data Source

4. In Search metric, search for an In-Memory DB metric name, such as redis_up, redis_memory_used_bytes, redis_connected_clients, redis_keyspace_hits_total, or redis_evicted_keys_total, to plot it as its own panel. Infrastructure metrics such as node_memory_MemAvailable_bytes and node_disk_io_time_seconds_total are also available from the same data source.

In-Memory DB cluster health metrics

5. Review the curated metric catalog in Metrics that includes infrastructure metrics and database health metrics; and view the corresponding metrics in Grafana.

6

Apply recommended alert thresholds

Configure the following alert rules in Grafana to detect the most common In-Memory DB failure modes early:

Alert	Threshold	What it indicates
Instance reachability	redis_up equal to 0	The instance is unreachable; alert immediately.
Key eviction	redis_evicted_keys_total increasing	Memory pressure causing keys to be evicted under the maxmemory policy; alert on any sustained increase.
Cache hit ratio	Below 90%	Rising cache misses that can indicate an undersized or cold cache.
CPU I/O wait	Above 1% (warning), above 2% (critical)	Storage bottleneck degrading command execution.
Available memory	Below 10% of node_memory_MemTotal_bytes	Node-level memory pressure that can affect the instance.
Filesystem free space	Below 15% of node_filesystem_size_bytes	Risk of write failures on persistence volumes.
Disk I/O saturation	rate(node_disk_io_time_seconds_total[5m]) above 0.9	I/O saturation impacting persistence and snapshotting.

Example PromQL for the cache hit ratio alert:

rate(redis_keyspace_hits_total[5m])
/
(
  rate(redis_keyspace_hits_total[5m])
  + rate(redis_keyspace_misses_total[5m])
)

Example PromQL for the available memory alert:

(node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes) * 100 < 10

Result: The metrics and logs for In-Memory DB v2 clusters are successfully available in the Grafana dashboard.

Troubleshooting

No data appears in Grafana after activation

Both layers must report true before data flows. Verify that the contract-level Logging and Monitoring services are active for the region and that the cluster properties block shows logsEnabled: true and metricsEnabled: true. If either layer is false, no data flows regardless of the other layer's state.

API accepts logsEnabled: true but no data flows

This is an expected behaviour. If you set logsEnabled or metricsEnabled to true while the contract-level service is inactive, the API accepts the request without error. Activate the corresponding Logging or Monitoring Service for the region first, then verify that the cluster property is still set to true.

Grafana sign-in fails or shows a permission error

Only administrator users and users who have been explicitly granted access to the observability product can sign in. If you have never previously interacted with the observability product, your permissions may not have propagated yet. Wait a few minutes and try again, or ask your contract administrator to confirm that observability access has been granted.

Activation response does not include a grafanaEndpoint

A missing grafanaEndpoint in the activation response indicates that the request body was malformed or that the service was not activated successfully. Confirm that the request body includes "enabled": true inside a valid properties object and that the Content-Type: application/json header is present. Re-send the request and check the response for an enabled: true field alongside the endpoint URL.

Searching for a metric name returns no results

In-Memory DB v2 database health metrics use the redis_ prefix (for example, redis_up, redis_connected_clients). If a metric name returns no results, confirm you are searching for a redis_-prefixed name. Infrastructure metrics use the node_ prefix.

Conclusion

You configured end-to-end observability for an In-Memory DB v2 cluster on IONOS CLOUD. The two-layer activation model gives you contract-level control over billing and per-cluster control over telemetry granularity. The curated metric catalog and recommended thresholds provide a starting point for production-grade alerting.

Next steps

• Extend the default dashboards with custom PromQL panels for command latency and connection saturation.
• Explore the In-Memory DB v2 API specification for more cluster properties such as eviction policy tuning and snapshot scheduling.