[RELEASE] Scylla Monitoring Stack 4.12.1

Release Notes
1

The ScyllaDB team is pleased to announce the release of ScyllaDB Monitoring Stack 4.12.0 + 4.12.1

Due to an issue with 4.12.0 we jump to 4.12.1

ScyllaDB Monitoring Stack is an open-source stack for monitoring ScyllaDB based on Prometheus and Grafana. ScyllaDB Monitoring Stack 4.12.1 supports:

  • ScyllaDB 2024.1, 2025.1, 2025.2, 2025.3, and the upcoming 2025.4 release

  • ScyllaDB Manager 3.x

This release introduces an updated look to the overview dashboard and a new dashboard for the upcoming vector search.

Related Links

Version updates for ScyllaDB Monitoring Stack 4.12.0

  • Prometheus upgraded to version 3.6.0

  • Grafana upgraded to version 12.2.0

New Information in ScyllaDB Dashboards

Overview Dashboard Change

  • The Overview dashboard received a general facelift to adapt to non-heterogeneous node sizes and x-cloud.

[Upload failed]

  • Cluster status is now displayed in a panel showing the cluster status in text and a panel that shows the number of active nodes (and if any: joining, leaving, and unreachable). #2593

  • The new service groups table summarizes the available service groups, and act as a quick navigation #2592

[Upload failed]

  • A new Data center summarizes the Different data centers and act as a quick navigation.

image
image710×169 9.94 KB

  • The top stats panel shows information over time #2595

[Upload failed]

  • The number of cores has been added to the node table #2591

  • The DC disk usage graph shows all the nodes in (vs average) that DC #2666

  • The manager panel was removed from the top row #2594

  • The cluster name is now shown (not just ID) when available #2589

Vector Search dashboard

  • The upcoming ScyllaDB Vector search service runs on a dedicated node. The new vector-search dashboard monitors both the vector search service and the node physical statistics.

  • Node metrics include: Disk, Memory and CPU usage

  • General status indicates if it’s available or not

  • Per index information combines:

    • The index size (number of vectors in the index)

    • Index build rate (how many new vectors per second are added to the index)

    • Index requests rate

    • Request latencies in Average time, P95, and P99

[Upload failed]

Detailed Dashboard Change

  • Add foreground Read and Write per service group #2652
    [Upload failed]

Keyspace Dashboard change

  • ScyllaDB 2025.4 introduces per-index statistics about Index usage. A new row in the keyspace dashboard shows that per-index statistic #2669
    image
    image1600×241 70.2 KB

OS Dashboard Change

  • Switch network traffic panels to the more comfortable bps units #2641

Bug Fixes

  • Tablets load balancer, make the title and description clearer #2640

  • LWT panels are not displayed well (full label data makes them hard to read). #2637

  • Write AWait panel - tool tip is not formatted properly. #2625

  • Update the hint panel to the new metric name #2619

  • The -c option in start-all.sh does not support values with spaces #2602

  • The device drop-down in the OS Dashboard should only show node_exporter devices #2628

Operational Changes

  • Limit the SG dropdown (Service Level Group) in the Detailed dashboard to: user-facing service groups #2511

  • When adding the vector-search service, add the node_exporter targets for that node #2636

  • Read Grafana server metrics, enable alerts on Grafana server failure #2632

  • Allow Vector service to work with Prometheus directory binding #2667

  • Documentation: The docker-compose.example.yml is automatically updated on every release #2627

  • Documentation: Add usage examples for the -c option in start-all.sh #2601

  • Enable experimental-promql-functions for Prometheus #2629