Qdrant Integration
Monitor collection point counts, search latency, and segment optimization metrics across your Qdrant instances. Get per-collection SLO tracking and AI segment advisor recommendations for your vector search engine.
How It Works
Enable Qdrant Telemetry
Enable the built-in Prometheus metrics endpoint in your Qdrant service config or via the QDRANT__SERVICE__ENABLE_METRICS=true environment variable. The metrics endpoint exposes search latency histograms, point counts, and segment statistics.
Deploy TigerOps Agent
Install the TigerOps agent alongside your Qdrant instance. It scrapes the /metrics endpoint and auto-discovers all collections, their vector configurations, and shard topology for cluster deployments.
Configure Per-Collection Alerts
TigerOps creates per-collection dashboards for search latency, indexing throughput, and segment health. Define p99 latency SLOs and segment optimization lag thresholds per collection independently.
Monitor Distributed Sharding
For Qdrant cluster deployments, TigerOps tracks shard distribution across nodes, peer replication state, and consensus health so you can detect imbalanced shards and replication failures early.
What You Get Out of the Box
Collection Point Count Tracking
Per-collection indexed and indexed-deleted point counts, vectors_count vs. points_count delta, and point growth rate tracking to monitor your vector data population over time.
Search Latency Monitoring
Per-collection and per-shard search latency at p50, p95, and p99 with search request rates, timeout rates, and query type breakdown for nearest neighbor vs. filtered search operations.
Segment Optimization Metrics
Segment count per collection, optimization task queue depth, merge operation progress, indexing status of unindexed vectors, and HNSW graph construction completion rates.
Cluster Shard Health
Per-shard point distribution, shard replication factor compliance, remote shard availability, and Raft consensus state across all nodes in your distributed Qdrant cluster.
Memory & Storage Utilization
Per-collection memory usage for in-memory and memmap vectors, disk usage for on-disk indexes, and mmap page fault rates that indicate index segments being loaded from disk.
AI Segment Optimization Advisor
TigerOps AI detects when unoptimized segments are causing search latency regressions, recommends optimal optimizer configuration parameters, and predicts when optimization will complete.
TigerOps Agent for Qdrant
Enable Prometheus metrics in Qdrant and configure the TigerOps agent to collect and forward them.
# Qdrant service config (config.yaml)
# service:
# enable_metrics: true
# metrics_port: 6333 # same as HTTP port, path /metrics
# TigerOps Qdrant Agent Configuration
receivers:
qdrant_prometheus:
endpoints:
- url: "http://qdrant-node-0:6333/metrics"
node: "node-0"
- url: "http://qdrant-node-1:6333/metrics"
node: "node-1"
- url: "http://qdrant-node-2:6333/metrics"
node: "node-2"
collection_interval: 15s
qdrant_api:
endpoint: "http://qdrant:6333"
api_key_env: QDRANT_API_KEY
# Per-collection stats polling
collections_poll_interval: 30s
# Collection-specific SLOs
collection_slos:
- name: "documents"
search_latency_p99_ms: 50
max_unindexed_vector_ratio: 0.1
- name: "images"
search_latency_p99_ms: 100
max_segment_count: 20
exporters:
tigerops:
endpoint: "https://ingest.atatus.net/api/v1/write"
bearer_token: "${TIGEROPS_API_KEY}"
send_interval: 15s
alerts:
optimizer_queue_depth: 10
shard_replication_factor_compliance: true
memory_usage_gb: 16
search_timeout_rate_pct: 1Common Questions
Which Qdrant versions does TigerOps support?
TigerOps supports Qdrant 1.0 and later via the built-in Prometheus metrics endpoint. Both single-node and distributed cluster modes are supported. Qdrant Cloud managed deployments are also supported via the Qdrant REST API metrics polling mode.
How does TigerOps track Qdrant segment optimization progress?
TigerOps monitors the qdrant_collection_optimizer_status metric and the segment count metrics to detect when collections have large numbers of unoptimized segments. Alerts fire when the unindexed vector ratio exceeds your configured threshold.
Can TigerOps monitor Qdrant quantization effectiveness?
Yes. TigerOps tracks quantization-related metrics including the ratio of quantized vs. original vectors stored, search accuracy scores (when using oversampling), and memory savings achieved from scalar or product quantization configurations.
How does TigerOps handle Qdrant cluster resharding events?
TigerOps monitors the Qdrant cluster API for shard transfer operations and resharding progress. During resharding, TigerOps annotates the metric timeline and sends alerts if a shard transfer stalls or if point counts diverge between source and target shards.
Does TigerOps support monitoring Qdrant payload index performance?
Yes. TigerOps tracks payload index build times, indexed payload field cardinalities, and filtered search performance metrics. When filtered searches are slower than vector searches, TigerOps suggests which payload fields would benefit from keyword or integer indexing.
Stop Discovering Qdrant Segment Bloat After Search Performance Degrades
Segment optimization monitoring, search latency SLOs, and AI advisor recommendations. Deploy in 5 minutes.