greptimedb/tests/perf

Query performance regression harness

This directory is for query performance cases that compare a base build with a candidate build. It is not a replacement for sqlness: the goal is to measure the effect of optimizer/query-engine changes on realistic scan work.

The Prometheus remote-write cases are end-to-end regression coverage for the automatic write, flush, SST, and query path. They do not replace controlled encoding experiments that explicitly compare plain/dictionary, no_dictionary, BYTE_STREAM_SPLIT, or Auto policies.

Phase 1: direct readable SST fixtures

Phase 1 should generate data by writing readable Mito SST files and matching manifest checkpoints directly. This follows the gc_readable_sst_fixture lab approach from ~/greptimedb-gc-huge-stress: use Mito's SST writer to create queryable files, then write a checkpoint and _last_checkpoint that reference those files.

The generator itself must be generic. It should not know about a specific issue such as #7913 or a specific PromQL query. Cases provide declarative table schema, data layout, distributions, and queries; the generator turns those declarations into readable SST fixtures.

The intended flow for each case is:

1. Start a GreptimeDB build and create an empty table to seed catalog/table metadata.
2. Stop the process.
3. Use the seed region metadata/manifest to generate deterministic readable SSTs and a replacement manifest checkpoint offline.
4. Start the same build on the generated data directory.
5. Run warmup and measured queries.
6. Repeat the same fixture/query process for the candidate build.
7. Compare base vs candidate metrics and write a regression report.

Direct SST fixtures are the default for phase 1 because they provide stable file counts, time ranges, row groups, and label distributions without spending CI time on ingestion and flush. Ingestion-path cases can be added later for nightly or release-level realism.

Prometheus remote-write scenario

The runner also supports scenario.kind = "prom_remote_write_then_query" for a bounded write-path smoke/regression flow. This path is explicit and separate from the direct-SST fixture path: it starts the base and candidate distributed clusters, writes deterministic Prometheus remote-write v1 samples through /v1/prometheus/write, flushes the configured physical metric table, checks visibility, runs the configured SQL/TQL queries, stops and awaits the datanode, then optionally inspects the landed SST/Parquet footer, encoding, and size and runs the read-bench against the quiescent data directory.

Remote-write cases configure one database, one logical metric, and one physical table under [scenario.remote_write]:

[scenario]
kind = "prom_remote_write_then_query"

[scenario.remote_write]
database = "public"
metric = "prom_remote_write_seeded_random"
physical_table = "greptime_physical_table"
series_count = 512
samples_per_series = 1440
sample_chunk_size = 480
flush_every_sample_chunks = 1
start_unix_millis = 1_704_067_200_000
step_millis = 60_000
chunk_series_count = 128
timeout_seconds = 180
visibility_timeout_seconds = 120
# Optional: split by time so each helper invocation writes only this many samples
# per series, then periodically flush the physical metric table to produce
# multiple time-interval SSTs.

[scenario.remote_write.prom_store]
pending_rows_flush_interval = "1s"
max_batch_rows = 100000

Cases can optionally add [scenario.remote_write.value] to control the generated sample values without changing label cardinality or the runner lifecycle:

[scenario.remote_write.value]
pattern = "quantized_signal" # linear, constant, modulo, unique, seeded_random,
                              # run_length, quantized_signal,
                              # signal_with_sporadic_stalls, mixed_signal_repeated
base = 0.0
step = 0.125
cardinality = 4096           # buckets for modulo/seeded_random/quantized_signal/run_length
seed = 12345                 # deterministic seeded_random input
run_length = 8               # adjacent samples per bucket for run_length/quantized_signal
stall_every = 100            # interval for signal_with_sporadic_stalls
stall_length = 16            # held samples inside each stall interval
mixed_every = 5              # every Nth sample becomes the repeated base value

The default linear pattern preserves the helper's historical formula. Use constant or low-cardinality modulo/seeded_random values for repeated-value data shapes, run_length for run-heavy low-cardinality series, quantized_signal for signal-like values collapsed into a finite bucket set, signal_with_sporadic_stalls for mostly continuous signals with periodic flat spots, and mixed_signal_repeated for signal-plus-periodic-default mixtures. unique or high-cardinality buckets still work for broad sample-value distributions. This is a generic sample-value control for query/ingestion cases; it does not inspect or assert storage encoding, Parquet footers, or storage policy choices. For chunked remote-write ingestion, the runner passes the sample offset and total sample count to the helper so non-linear value patterns use a stable global/per-series ordinal across chunks.

Case schema, value distribution defaults, storage defaults, and read-bench defaults are owned by Rust. The Python runner calls query_perf_fixture plan --case <case.toml> and orchestrates the normalized JSON. The same helper exposes direct-sst, prom-remote-write, and inspect-footer subcommands; the old direct invocation (query_perf_fixture --case ... --out-dir ...) remains compatible.

Remote-write cases that need to validate storage output can add [scenario.remote_write.storage]. When present, the scenario body becomes: remote-write → flush → visibility check → query measurement → stop and await datanode → Parquet footer/size/encoding inspection → optional read-bench. Footer inspection is handled by query_perf_fixture inspect-footer, which reads local Parquet footers directly; it does not require Python pyarrow.

[scenario.remote_write.storage]
inspect = true
column = "greptime_value"
include_metadata_files = false
# Optional: inspect below datanode data home instead of the whole fresh data dir.
# root_suffix = "greptime/public/<table_id>"
min_files = 1
min_files_with_column = 1
require_encodings = ["BYTE_STREAM_SPLIT"]
forbid_encodings = ["PLAIN_DICTIONARY"]
max_total_file_size_bytes = 104857600
max_column_compressed_size_bytes = 52428800
max_column_uncompressed_size_bytes = 209715200
max_candidate_total_file_size_regression_pct = 10.0
max_candidate_column_compressed_size_regression_pct = 10.0
max_candidate_column_uncompressed_size_regression_pct = 10.0

The *_pct storage thresholds are percent regression limits comparing candidate against base; pct means percent, not percentile. Per-target storage checks (min_files, min_files_with_column, required/forbidden encodings, and absolute max_*_bytes) run for both base and candidate. Comparative max_candidate_*_regression_pct checks remain base-vs-candidate. These checks are generic footer and byte-size assertions and do not encode product-specific heuristics. When storage inspection is enabled, min_files and min_files_with_column default to 1 even if omitted, so dry-runs show these planned checks and an empty flush or missing target column fails the scenario. By default the inspector root is the target datanode data home, which is suitable for a fresh single-case data directory. For reused or more complex data homes, set root_suffix to a path relative to the datanode data home to narrow inspection.

When storage inspection is enabled, [scenario.remote_write.read_bench] defaults to enabled with both datanode parquetbench and scanbench. Disable it with enabled = false. parquetbench measures per-SST reader cost, scanbench measures region scan cost, and query measurements still exercise the SQL/TQL frontend path. Treat all performance conclusions as release-only; debug builds are suitable only for command wiring and correctness checks.

The runner creates the configured database if needed, writes a per-target frontend config enabling [prom_store] with metric engine storage and a non-zero pending_rows_flush_interval, and validates that the logical metric table reaches series_count * samples_per_series rows before trusting the query measurements. Use --fixture-generator /path/to/query_perf_fixture to provide the Rust helper. Deprecated --remote-write-generator and --storage-inspector options are kept only for CLI compatibility. --fixture-only is rejected for remote-write cases; use --dry-run for planning.

Large manual remote-write cases can set sample_chunk_size to split ingestion by time. For each chunk, the runner invokes query_perf_fixture prom-remote-write with the same series cardinality but a shorter --samples-per-series and an advanced --start-unix-millis. flush_every_sample_chunks controls periodic ADMIN FLUSH_TABLE('<physical_table>') calls; with flush_every_sample_chunks = 1, each time chunk is flushed separately. The final visible SST/file-range layout is still determined by the storage engine's normal compaction policy, so cases that need multi-window file distribution should span multiple compaction windows. If sample_chunk_size is omitted, the runner keeps the older single-helper-invocation behavior and flushes once at the end.

The default remote-write coverage set contains four cases with the same 2048 series × 14,400 samples (29,491,200 rows per target) shape. Each writes ten one-day chunks, flushes every chunk, requires at least two visible SSTs with the value column, runs seven read-bench iterations with parquetbench capped at four SSTs while scanbench covers the landed region, and measures one ten-day TQL selector with two warmups and 15 iterations:

• prom_remote_write_seeded_random: high-distinct seeded-random values with cardinality 29,491,200 and seed 8444.
• prom_remote_write_run_heavy: low-cardinality values in exact 16-sample runs.
• prom_remote_write_mixed_every: continuous signal values with every fifth sample repeated at the base value.
• prom_remote_write_integer_counter: strictly increasing integer counter values, with disjoint ranges for each series.

These cases record normal storage/footer and read-bench results but do not gate specific value encodings or storage-size outcomes. Use the controlled encoding experiment matrix for those policy comparisons.

tests/perf/query_cases/prom_remote_write_7913/case.toml is a larger manual case for issue #7913. It writes 8192 series × 20160 samples through remote-write in 1440-sample daily time chunks, flushing after each chunk before running 1d/7d/14d TQL selectors. It is not included in the default all case set because ingestion cost dominates routine CI validation. Adding the heavy-regression PR label runs only this case; query-regression runs the five routine default cases. Manual workflow dispatch accepts the heavy token to select this case.

OTLP trace load scenario

scenario.kind = "otlp_trace_load" runs a bounded native otelgen process against each local distributed cluster. The runner excludes the configured warmup window, derives throughput and mean request latency from GreptimeDB's OTLP-specific metrics, flushes the trace table, and verifies its row count against the accepted-span counter. The case is intentionally outside the default set until its variance is known.

Build base/candidate greptime binaries with the same profile, build the candidate query_perf_fixture, then run the case explicitly:

WORK_DIR="$(mktemp -d /tmp/query-perf-otlp.XXXXXX)"
REPORT="$WORK_DIR/trace-report.json"
python3 tests/perf/query_regression_runner.py \
  --case tests/perf/query_cases/otlp_trace_load/case.toml \
  --base-bin /path/to/base/target/nightly/greptime \
  --candidate-bin /path/to/candidate/target/nightly/greptime \
  --fixture-generator /path/to/candidate/target/nightly/query_perf_fixture \
  --otelgen-bin /path/to/otelgen \
  --work-dir "$WORK_DIR" \
  --output "$REPORT"

Install YouPlot once, then render all comparison metrics and threshold results in the terminal:

brew install youplot
tests/perf/plot_otlp_trace_report.sh "$REPORT"

For each target, accepted_spans should equal table_rows, and failures should stay within max_failure_count. Throughput is better when spans_per_second is higher; its actual_pct is (base - candidate) / base * 100. Latency is better when mean_latency_ms is lower; its actual_pct is (candidate - base) / base * 100. A positive actual_pct is a candidate regression, while a negative value is an improvement. The case passes when every actual_pct is at or below its limit_pct and every failure-count threshold passes. For local results, run the case at least three times on an otherwise idle machine and compare the median regressions rather than relying on one run.

The CI runner image includes the pinned otelgen binary. Until this case is added to the default set, run it explicitly with workflow_dispatch:

gh workflow run query-regression.yml \
  --ref <workflow-branch> \
  -f case=tests/perf/query_cases/otlp_trace_load/case.toml \
  -f base_ref=<full-base-sha> \
  -f candidate_ref=<full-candidate-sha> \
  -f cargo_profile=nightly \
  -f http_timeout=300 \
  -f runner=perf-regression-8-cores

The selected ARC scale set must already be deployed with the runner-image digest built from the current query-regression Dockerfile.

Generator contract

The direct-SST generator should accept a case definition with:

• one or more table definitions: columns, semantic types, primary key, time index, SST format, append mode
• deterministic distributions: seed, series/tag cardinalities, label/value functions, timestamp layout
• physical layout: regions, SST count, rows per SST, row group size, time ranges per SST, optional overlap/skew
• output paths for object-store files, manifest checkpoints, and fixture metadata

This keeps query regression cases reusable: the same generator can produce PromQL, SQL, pruning, projection, join, or aggregation fixtures by changing only case config.

What a case owns

Each optimization PR should add or update the query case for the pattern it is expected to affect. A case should define:

• schema and seed table SQL
• deterministic data shape: seed, series count, rows per SST, SST count, time range layout, label distribution, region/partition layout
• queries to run
• warmup/measurement repetitions
• metrics to collect
• base-vs-candidate thresholds

The [case] table is metadata for reports. The executable regression config lives under [scenario]. A scenario owns data generation, queries, and thresholds:

[case]
name = "example"
description = "what this regression protects"

[scenario]
kind = "direct_readable_sst"
seed = 12345

[[scenario.tables]]
# table schema and distributions

[scenario.layout]
# SST and series layout

[[scenario.queries]]
# query, warmups, iterations, thresholds

The runner currently supports direct_readable_sst and prom_remote_write_then_query.

Metrics

Primary gates should compare query work rather than plan text:

• scanned files / file ranges
• scanned rows or row groups
• bytes read when available
• pruning ratio
• query latency median/p95
• output row count as a sanity check

Plan details such as pushed filters are useful diagnostics, but should not be the main pass/fail signal.

Runner MVP

query_regression_runner.py is the base-vs-candidate orchestration layer. The current MVP parses a case, creates per-target work directories, and in real query mode starts a local distributed cluster for each target: metasrv (memory-store, region failover disabled), one datanode (node_id=0), and one frontend. It creates the configured Mito table(s) through frontend HTTP SQL, discovers the real one-region-per-table metadata via information_schema, stops only the owning datanode, generates one shared direct-SST fixture per table using the discovered --region-id, --table-dir, and --table, injects those region subtrees into the datanode data home, restarts the datanode, then validates and measures through frontend. Reports are written as JSON under the work directory.

The runner intentionally keeps metasrv alive for the whole target run because memory-store metadata would otherwise be lost. It replaces only the discovered datanode region directory under data/greptime/<schema>/<table_id>/... with generated SST files and a manifest checkpoint. For multi-table cases this is repeated per table, enabling true JOIN fixtures while still requiring exactly one region per table. Base and candidate must discover identical per-table table_dir and region_id; otherwise the run fails.

Multi-table direct-SST cases must use unique table names as well as unique (database, name) pairs because the generator currently selects a table with --table <name>. Per-table fixture directories are derived from table index, database, and table name with path-unsafe characters sanitized.

Currently enforced threshold:

• max_candidate_latency_regression_pct, based on client-side median latency.

Server-side scan thresholds such as file ranges and scanned rows are planned for a follow-up PR that extracts them from structured EXPLAIN ANALYZE VERBOSE output. Do not add those threshold keys until the runner enforces them.

Dry-run example:

uv run --no-project python tests/perf/query_regression_runner.py \
  --case tests/perf/query_cases/promql_pushdown_7913/case.toml \
  --base-bin /path/to/base/greptime \
  --candidate-bin /path/to/candidate/greptime \
  --work-dir /tmp/query-perf-work \
  --dry-run

With a fixture generator:

uv run --no-project python tests/perf/query_regression_runner.py \
  --case tests/perf/query_cases/promql_pushdown_7913/case.toml \
  --base-bin /path/to/base/greptime \
  --candidate-bin /path/to/candidate/greptime \
  --fixture-generator /path/to/query_perf_fixture \
  --fixture-cache-dir /mnt/query-regression-fixtures \
  --allow-large-fixture \
  --work-dir /tmp/query-perf-work

This mode launches metasrv, datanode, and frontend for each target with explicit localhost HTTP/gRPC/MySQL/Postgres ports and writes component stdout/stderr under each target's logs/ directory.

By default query mode requires fresh base/candidate work directories and fails if either target directory already exists with contents. Use --reuse-work-dir only when intentionally debugging an existing run directory. SQL HTTP requests default to a 120 second timeout; override with --http-timeout <seconds> for slow lab runs.

For large direct-SST fixtures, pass --fixture-cache-dir <dir> to store generated fixtures in a persistent content-addressed cache keyed by case name and fixture data configuration. Query and threshold edits reuse the same cached data as long as the scenario layout and table definitions do not change. Cached fixtures are reused automatically when their summary.json matches the discovered table/region metadata; incompatible entries are regenerated instead of reused. Fixture materialization keeps base and candidate data directories isolated, but copies files efficiently by trying filesystem reflinks first, hardlinks for immutable SST/object files next, and normal copies as a fallback. Manifest files are reflinked or copied, not hardlinked.

Fixture generator smoke test:

cargo run -p cmd --bin query_perf_fixture --features dev-tools -- \
  direct-sst \
  --case tests/perf/query_cases/smoke_direct_sst/case.toml \
  --out-dir /tmp/query-perf-smoke

Runner smoke test with fixture generation only:

uv run --no-project python tests/perf/query_regression_runner.py \
  --case tests/perf/query_cases/smoke_direct_sst/case.toml \
  --base-bin /path/to/query_perf_fixture \
  --candidate-bin /path/to/query_perf_fixture \
  --fixture-generator /path/to/query_perf_fixture \
  --work-dir /tmp/query-perf-runner-smoke \
  --fixture-only

--fixture-only preserves the earlier smoke behavior: it does not start standalone servers, and it materializes the generated fixture into base and candidate data directories for plumbing validation.

Remote-write runner dry-run:

uv run --no-project python tests/perf/query_regression_runner.py \
  --case tests/perf/query_cases/prom_remote_write_seeded_random/case.toml \
  --base-bin /path/to/base/greptime \
  --candidate-bin /path/to/candidate/greptime \
  --fixture-generator /path/to/query_perf_fixture \
  --work-dir /tmp/query-perf-remote-write \
  --dry-run

GitHub Actions

.github/workflows/query-regression.yml provides an opt-in CI entrypoint for query regression runs. It builds its own binaries for now:

• base greptime from the PR base commit, or workflow_dispatch base_ref
• candidate greptime and query_perf_fixture from the PR merge ref/current candidate checkout
• runner and summary formatter from the candidate checkout

The workflow builds base and candidate greptime as normal release-equivalent binaries. Candidate query_perf_fixture is the extra head-side helper binary; the runner uses candidate greptime datanode parquetbench/scanbench as the read-bench tool against each target's data directory.

The workflow runs automatically only when query-regression or heavy-regression is added to a non-draft PR; it does not rerun on pushes, ready-for-review, or reopen events. query-regression runs the five routine default cases, while heavy-regression runs only the high-cardinality remote-write #7913 case. PR runs build base/candidate once and use --allow-large-fixture. Manual workflow_dispatch runs can pass all, heavy, one case path, or a comma/whitespace-separated list of case paths, and can override refs. The main report artifact uploads only aggregate/per-target JSON reports, component logs, and query-regression-summary.md with seven-day retention; fixture data, SSTs, and cluster state are excluded. PR runs also upload a separate trusted-comment artifact containing PR metadata and aggregate reports. The workflow writes the Markdown summary to the workflow step summary and updates a sticky PR comment through the trusted follow-up workflow.

Built-in cases

The promql_pushdown_7913 case is only one case using the generic fixture format. It generates a high-cardinality metric-like table with a nanosecond time index and many SSTs with non-overlapping time ranges. Its timestamp_major series layout writes one sample for every series at each scrape timestamp, so short PromQL/TQL selector windows still scan realistic raw sample volumes. The queries should show scan-level time filters, tight SST pruning, and enough raw rows to make distributed PromQL pipeline placement meaningful instead of a millisecond-scale canary.

Additional SQL optimizer cases:

• sql_topk_order_by: single-table TopK / ORDER BY on a DOUBLE field with time and tag predicates.
• sql_aggregate_order_by: grouped aggregate ordered by aggregate value with a LIMIT.
• sql_join_filter_order: two direct-SST tables joined on a shared tag with time filters, aggregate ordering, and LIMIT.