All Integrations
DatabasesTigerOps agent

SingleStore Integration

Monitor pipeline ingestion rates, columnstore segment metrics, and leaf node health across your SingleStore deployments. Get AI-powered ingestion anomaly detection and resource manager visibility.

Setup

How It Works

01

Install TigerOps Agent on Aggregator

Deploy the TigerOps agent on your SingleStore aggregator node. The agent connects to the information_schema and sys.* tables to collect cluster topology, leaf node health, and resource manager metrics.

02

Configure Monitoring Credentials

Create a SingleStore monitoring user with SELECT on information_schema.*, sys.*, and the PROCESS privilege. The agent uses this user to collect query stats, pipeline status, and columnstore segment metrics.

03

Enable Pipeline and Columnstore Monitoring

TigerOps queries information_schema.PIPELINES and PIPELINE_BATCHES to track ingestion rates, batch errors, and lag per pipeline. Columnstore metrics are pulled from SYSINDEXSTATISTICS and SYSSEGMENTS.

04

Set Ingestion and Leaf Node Alerts

Define thresholds for pipeline batch error rates, leaf node disk saturation, and query concurrency limits. TigerOps correlates leaf node failures with query routing errors and pipeline ingestion drops.

Capabilities

What You Get Out of the Box

Pipeline Ingestion Monitoring

Per-pipeline batch ingestion rate, rows ingested per second, batch error counts, and pipeline lag. TigerOps alerts when ingestion drops below baseline or batch errors exceed your error rate SLO.

Columnstore Segment Metrics

Segment count per table, segment size distribution, merge queue depth, and columnstore memory pressure. Detect when segment accumulation is degrading query scan performance.

Leaf Node Health

Per-leaf disk usage, memory utilization, CPU load, and connectivity status from the aggregator perspective. TigerOps alerts when a leaf node is unreachable or approaching resource limits.

Query Performance Analysis

Top queries by execution time, memory usage, and rows examined from information_schema.PROCESSLIST and SYSQUERIES. Identify which concurrent queries are consuming the most leaf node resources.

Resource Manager Utilization

Resource pool CPU and memory quotas, active workload per resource group, and throttled query counts. TigerOps tracks whether resource manager limits are causing query queuing.

AI Ingestion Anomaly Detection

TigerOps AI baselines your SingleStore pipeline ingestion patterns and fires predictive alerts when ingestion rates deviate from expected patterns — catching upstream data source failures before they cause data gaps.

Configuration

TigerOps Agent Config for SingleStore

Connect the TigerOps agent to your SingleStore aggregator for pipeline, columnstore, and leaf node metrics.

tigerops-singlestore.yaml
# TigerOps SingleStore integration config
# Place at /etc/tigerops/conf.d/singlestore.yaml

integrations:
  - name: singlestore
    type: singlestore
    config:
      # Connect to aggregator node
      host: singlestore-aggregator.internal
      port: 3306
      user: tigerops_monitor
      password: "${SINGLESTORE_MONITOR_PASSWORD}"
      tls:
        enabled: true

      # Cluster monitoring
      cluster:
        collect_node_health: true    # MV_NODES table
        collect_resource_pools: true # RESOURCE_POOL table

      # Pipeline monitoring
      pipelines:
        enabled: true
        databases:
          - my_app_db
          - streaming_db
        collect_batch_errors: true
        batch_error_lookback_minutes: 5

      # Columnstore segment monitoring
      columnstore:
        enabled: true
        # Tables to monitor (empty = all columnstore tables)
        monitor_all_tables: true
        collect_segment_details: false  # high cardinality

      # Query performance
      query_stats:
        enabled: true
        collect_slow_queries: true
        slow_query_threshold_ms: 5000
        top_query_count: 20

    scrape_interval: 30s

remote_write:
  endpoint: https://ingest.atatus.net/api/v1/write
  bearer_token: "${TIGEROPS_API_KEY}"

# Monitoring user setup (run once on aggregator):
# CREATE USER 'tigerops_monitor'@'%' IDENTIFIED BY 'secure_password';
# GRANT SELECT ON information_schema.* TO 'tigerops_monitor'@'%';
# GRANT PROCESS ON *.* TO 'tigerops_monitor'@'%';
# GRANT SELECT ON my_app_db.* TO 'tigerops_monitor'@'%';
FAQ

Common Questions

Which SingleStore versions does TigerOps support?

TigerOps supports SingleStore DB (formerly MemSQL) versions 7.5 and later, including SingleStore Managed Service (cloud). The integration uses standard information_schema and sys.* tables available in all supported versions.

What privileges does the SingleStore monitoring user need?

The monitoring user needs SELECT on information_schema.*, SELECT on sys.*, and PROCESS global privilege. For pipeline monitoring, SELECT on the specific database containing the pipelines is also required. No DML or DDL privileges are needed.

How does TigerOps detect pipeline batch errors?

TigerOps queries information_schema.PIPELINE_BATCHES for error_count and error_message per pipeline batch. When batch errors exceed your configured threshold, TigerOps fires an alert with the error message and affected pipeline name, correlated with concurrent ingestion rate changes.

Can TigerOps monitor SingleStore Universal Storage tables alongside columnstore?

Yes. TigerOps collects metrics for both columnstore and rowstore table types from SYSINDEXSTATISTICS. Universal Storage tables that use a mix of both storage formats are tracked with separate metrics for their columnstore and rowstore segments.

How does TigerOps handle SingleStore leaf node failover events?

TigerOps monitors leaf node availability from the aggregator by checking INFORMATION_SCHEMA.MV_NODES. When a leaf transitions to an offline state, TigerOps fires an immediate alert with the leaf node identity and correlated query error rates observed during the failover window.

Get Started

Keep Your SingleStore Pipelines and Leaf Nodes Healthy

Pipeline ingestion monitoring, columnstore segment tracking, and AI anomaly detection. Deploy in 5 minutes.