Lei, HUANG
aa4abe554a
* feat(promql): add evaluated series math kernels Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * feat(promql): add evaluated series scalar arithmetic Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * feat(promql): complete evaluated series math functions Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * feat(promql): add series comparisons and scalar results Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * feat(promql): add sparse evaluated series basic aggregates Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * feat(promql): align time-varying scalar maps on evaluated grids Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): drop metric name in evaluated series math like Prometheus Prometheus' simpleFloatFunc drops __name__ together with __type__ and __unit__. Check abs, ceil, floor, sqrt and round against fixtures generated by the Prometheus v3.15.0 engine. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): drop empty labels with the metric name in series math Prometheus has no empty-valued labels, so a kept empty value would hide a same-labelset collision from the caller. Round shares its calculation with the UDF, so check it only against the Prometheus fixtures. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): align evaluated series map kernels with Prometheus Follow Prometheus v3.15.0 simpleFloatFunc, clamp, VectorscalarBinop and vectorElemBinop: drop __name__, __type__ and __unit__ for math, arithmetic and bool comparisons; use ports of Go math (including the amd64 Log and Exp assembly) for ^, %, atan2 and the transcendental functions; empty a clamp with max < min; keep -0 in sgn and Prometheus' operation order in deg and rad. Fields are compared independently and filtered output is trimmed with GridSeriesBuilder. Check the kernels against fixtures generated by the Prometheus engine and Go math results built for amd64. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): align evaluated series aggregation with Prometheus Follow Prometheus aggregation and generateGroupingLabels: no modifier is by (), without drops __name__, empty label values are absent, sum and avg use Neumaier compensation and avg switches to an incremental mean on overflow. Fields are aggregated independently, so a field without values at a row is null also for count and group. Output is trimmed with GridSeriesBuilder. Check against fixtures generated by the Prometheus engine and drop the main UDAF oracle and the common-function dependency. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): validate whole segments in series comparisons map_compare checked sample types one by one, so a histogram or mixed segment without a present sample passed. Validate segments up front as the other kernels do, check them before an empty clamp, share the empty-label rule with aggregation, and pin exact bits in the deg, rad and incremental avg tests. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * chore: exclude promql golden fixtures from typos Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * docs(promql): point series round at Prometheus funcRound Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): keep Go math ports only where std leaves the tolerance Series math now follows the relative-1e-10 comparison rule: functions use Rust's standard library unless Go's amd64 result differs beyond the tolerance or in overflow and sign. asin, atan, tanh, asinh, acosh and atanh use std; the Exp assembly port is reduced to its overflow edge, and sinh/cosh build on it above 21. Golden and Go math fixtures are compared with almost.Equal. The Go cases are evaluated through the kernels' own per-sample functions, and each retained port must have a case where std fails, so a port that is no longer needed, or a swap back to std, fails a test. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * refactor(promql): validate aggregate input like series maps aggregate_float repeated map's schema, grid-axis and float-segment checks in a different order, so one bad series could fail with a different error category in each. Both now use validate_float_series, and map_math/map_negate/map_arithmetic run it before mapping instead of their own axis check. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * test(promql): pin avg compensation to Prometheus bits With fixtures compared within the tolerance, dropping or misplacing the Kahan compensation in avg only moves the last bits. Three Prometheus cases where it does are compared exactly: compensation on the incremental mean, the separate division of sum and compensation, and rescaling the compensation when avg switches to the incremental mean. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * test(promql): pin the zero sign of clamp and the Neumaier swap Under the fixture tolerance, +0 and -0 compare equal and no case had an addend larger than the running sum, so two branches lost their tests: Go's Max picking +0 over -0, whose sign a later 1/x turns into the sign of an infinity, and the Neumaier swap, without which sum(1, 1e100, 1, -1e100) is 1 instead of 2. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * chore: allow the Unicode category Nd in typos Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * perf(promql): keep aggregate state in fixed-size grid blocks Aggregation kept one B-tree entry and one Vec allocation per populated evaluation point, plus a tree lookup per sample. State is now stored in 64-point blocks keyed by absolute grid index / 64, holding one state per field and point. A block is allocated only once a sample falls in it, so sparse input never expands to the full grid, and each field's run through a block costs one lookup. Points without samples are skipped on output, which also keeps the last block from reaching past the grid. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * perf(promql): drop reserved and empty labels in one rebuild drop_metric_name removed empty-valued labels one at a time, and each MetricName::remove copies the whole label array, so the cost grew with the square of the number of empty labels. Filter the labels once and rebuild the name with MetricName::try_new. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * test(promql): pin that aggregation keeps type and unit labels Prometheus generateGroupingLabels drops only the metric name, unlike math and arithmetic. Nothing failed when grouping dropped __type__ or __unit__. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * fix(promql): return aggregate groups in first-seen order Groups were collected in a BTreeMap and returned sorted by label. Prometheus' rangeEvalAgg appends groups in the order their first input appears, and a nested aggregation folds them in that order: for sum(sum by (g) (m)) over 1e308, -1e308, 1e308 with g = a, c, b, Prometheus gives 1e308 and label order overflows to +Inf. Look groups up by label key and keep them in a Vec. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> * test(promql): pin which zero min and max keep Prometheus replaces the min/max accumulator only on a strict comparison, so of +0 and -0 the first is kept. The golden tolerance treats the zeros as equal, and a non-strict comparison passed every test. Signed-off-by: Lei, HUANG <ratuthomm@gmail.com> --------- Signed-off-by: Lei, HUANG <ratuthomm@gmail.com>
Metrics, logs, and traces.
One engine, on your infrastructure.
A columnar database for metrics, logs, and traces on object storage. Apache-2.0 licensed core.
User Guide · API Docs · Roadmap 2026 · Slack
stable for production · canary includes pre-releases · nightly is a weekly snapshot of main
Introduction
GreptimeDB is an open-source observability database. Metrics, logs, and traces run on one columnar engine over object storage and share one table model: tags, timestamp, and fields. When signals carry common identifiers such as service, host, or trace ID, you can correlate them in SQL without moving data between databases.
Ingest through OpenTelemetry, Prometheus Remote Write, Loki Push, or Elasticsearch Bulk. Use SQL across observability data and PromQL for metrics. Migrate ingestion one signal at a time without rebuilding your collectors.
One Query Across Signals
OpenTelemetry ingestion writes spans to opentelemetry_traces and log records to
opentelemetry_logs. Both tables carry trace_id, so correlating them is a join:
-- The slowest failed spans in the last hour,
-- with the log lines emitted inside those same traces.
SELECT
t.service_name,
t.span_name,
t.duration_nano / 1000000 AS duration_ms,
l.timestamp AS log_time,
l.severity_text,
l.body
FROM opentelemetry_traces t
JOIN opentelemetry_logs l ON l.trace_id = t.trace_id
WHERE t.timestamp > now() - INTERVAL '1' HOUR
AND t.span_status_code = 'STATUS_CODE_ERROR'
ORDER BY t.duration_nano DESC
LIMIT 20;
Metrics join the same way, on any tag the tables share, such as service,
host, or pod.
Why You Might Use It
- You run Prometheus plus Loki or Elasticsearch and want one backend instead of three
- You have outgrown Prometheus on cardinality or retention and don't want the Thanos/Mimir operational surface
- You are hitting Loki's query performance limits as log volume grows
- You need long retention on object storage without a separate analytics stack
- You want to query telemetry with SQL, not only a domain query language
- You are storing GenAI or agent telemetry (OTel GenAI conventions) alongside infrastructure signals
Learn more in Why GreptimeDB.
What's Supported
| Ingest | OpenTelemetry (OTLP), Prometheus Remote Write, Loki Push, Elasticsearch Bulk, InfluxDB line protocol, gRPC |
| Query | SQL, PromQL, Jaeger-compatible trace queries, MySQL and PostgreSQL wire protocols |
| Storage | S3, GCS, Azure Blob and S3-compatible endpoints as primary storage, with memory and local-disk caches |
| Built in | Retention policies, downsampling, continuous aggregation, explicit table partitioning, and inverted / skipping / fulltext indexes |
Compute and storage are disaggregated: object storage holds the data, while memory and local-disk caches keep recent and frequently queried data close to compute.
Benchmarks
- Agent RCA Bench: LLM agents doing root cause analysis over GreptimeDB versus Prometheus + Loki + Tempo. 40% fewer wrong diagnoses, 48% fewer input tokens, 45% lower cost (write-up)
- GreptimeDB tops JSONBench's billion-record cold run test
- TSBS Benchmark
- More benchmark reports
Compatibility and Migration
Compatibility is per protocol, and query-side coverage is narrower than ingestion.
| Compatible | Not compatible | |
|---|---|---|
| Prometheus | Remote Write ingestion; PromQL queries | Gaps are listed in PromQL compatibility |
| Loki | Push ingestion; dual-write through Grafana Alloy makes the cutover gradual | LogQL and the rest of the Loki query API |
| Elasticsearch | _bulk ingestion in the open-source core; QueryDSL partially, in Enterprise |
Most other Elasticsearch APIs |
Limitations and Edition Boundary
Cluster deployment, object storage, the Flow engine, and every ingestion protocol listed above are in the Apache-2.0 build. Repartitioning, region migration, and index creation are manual operations there.
Read replicas, workload isolation, and automated repartitioning are GreptimeDB Enterprise features, along with enterprise security and governance. The Enterprise overview has the current list, and pricing has the edition comparison.
Architecture
GreptimeDB can run in two modes:
- Standalone — single binary for development and small deployments.
- Distributed — four components, each independently scalable:
- Frontend — protocol entry (OTel, Prometheus, MySQL/PostgreSQL, gRPC, ingestion APIs for Elasticsearch/InfluxDB/Loki) and the distributed query engine. Stateless, scales horizontally.
- Datanode — region engine with WAL, memtable, SST, cache, compaction, and indexes. Persists data to object storage. Elastic.
- Metasrv — metadata, routing, repartitioning, and security. Backed by a pluggable KV layer (etcd or RDS).
- Flownode (optional) — continuous flow computation (streaming and materialized views).
For deeper coverage, see the architecture doc or DeepWiki.
Try GreptimeDB
For AI agents — paste this prompt into your agent:
Read https://docs.greptime.com/SKILL.md and follow the instructions
to deploy, configure, ingest, and query GreptimeDB.
docker run -p 127.0.0.1:4000-4003:4000-4003 \
-v "$(pwd)/greptimedb_data:/greptimedb_data" \
--name greptime --rm \
greptime/greptimedb:latest standalone start \
--http-addr 0.0.0.0:4000 \
--grpc-bind-addr 0.0.0.0:4001 \
--mysql-addr 0.0.0.0:4002 \
--postgres-addr 0.0.0.0:4003
Dashboard: http://localhost:4000/dashboard
Read more in the full Install Guide.
Troubleshooting:
- Cannot connect to the database? Ensure that ports
4000,4001,4002, and4003are not blocked by a firewall or used by other services. - Failed to start? Check the container logs with
docker logs greptimefor further details.
Getting Started
Build From Source
Prerequisites:
- Rust toolchain — stable, pinned by
rust-toolchain.toml - Protobuf compiler (>= 3.15)
- C/C++ building essentials:
gcc/g++/autoconfand the glibc dev package (libc6-devon Ubuntu,glibc-develon Fedora) - Python toolchain (optional, only for some test scripts)
Build and run:
make # build greptime binary
cargo run -- standalone start # start in standalone mode
Common dev commands:
make fmt # format Rust code
make clippy # lint (fails on warnings)
make test # unit + integration tests (uses cargo-nextest)
make sqlness-test # SQL regression tests
See the Contribution Guidelines for the full developer workflow.
Tools & Extensions
- Kubernetes: GreptimeDB Operator
- Helm Charts: Greptime Helm Charts
- Dashboard: Web UI
- gRPC Ingester: Go, Java, C++, Erlang, Rust, .NET, TypeScript
- Grafana Data Source: GreptimeDB Grafana data source plugin
- Grafana Dashboard: Official Dashboard for monitoring
Project Status
GreptimeDB is generally available, with stable APIs and regular releases. It runs in production at scale — OceanBase Cloud operates 80+ GreptimeDB clusters managing 300 TB of logs, cutting log storage cost by 60%+ after migrating from Grafana Loki. See more in case studies.
Release lines and support windows are in the version reference. For where the project is going, read the v1.0 highlights and the 2026 roadmap.
Community
We invite you to engage and contribute!
If GreptimeDB is useful to you, please star the repo.
License
GreptimeDB is an open-core project. Its core is licensed under the Apache License 2.0.
A small set of peripheral, enterprise-only features are gated behind the
enterprise Cargo feature (not built by default) and are governed by the
separate GreptimeDB Enterprise License. Source files under
that license carry an explicit Enterprise License header.
Commercial Support
Scaling observability on your infrastructure? GreptimeDB Enterprise adds the operational, security, and support layer for production deployments. Contact us for details.
Contributing
- Read our Contribution Guidelines.
- Explore Internal Concepts and DeepWiki.
- Pick up a good first issue and join the #contributors Slack channel.
Acknowledgement
Special thanks to all contributors! See AUTHOR.md.
- Uses Apache Arrow™ (memory model)
- Apache Parquet™ (file storage)
- Apache DataFusion™ (query engine)
- Apache OpenDAL™ (data access abstraction)
All trademarks, logos, and brand names referenced in this README and in the Overview diagram are the property of their respective owners. Their use is for identification purposes only and does not imply endorsement or affiliation.