Skip to content

[pull] main from SigNoz:main - #943

Merged
pull[bot] merged 1 commit into
code:mainfrom
SigNoz:main
Aug 29, 2026
Merged

[pull] main from SigNoz:main#943
pull[bot] merged 1 commit into
code:mainfrom
SigNoz:main

Conversation

@pull

@pull pull Bot commented Aug 29, 2026

Copy link
Copy Markdown

See Commits and Changes for more details.


Created by pull[bot] (v2.0.0-alpha.4)

Can you help keep this open source service alive? 💖 Please sponsor : )

… segfault (#12720)

#### Description

- A PromQL subquery without a step, for example
`max_over_time(metric[5m:])`, segfaulted the whole query-service. The
engine calls `NoStepSubqueryIntervalFn` for such subqueries, and we
build the engine without it, so the call hits a nil function.
- The bug is present on every PromQL surface, because all of them share
the one engine constructor in `pkg/prometheus/engine.go`: v3 and v5
`query_range`, `/api/v1/query`, the clickhousev2 transpiler, and promql
alert rules. A saved rule with such a subquery crash-loops the instance
on its own schedule.
- The fix sets the callback to 1m. This matches the Prometheus default
global `evaluation_interval`, which upstream wires into this field. One
place fixes every path.
- This is the root cause of the SigNoz/platform-pod#3068 incident. The
instance-hardening request from that incident is tracked in
SigNoz/pulse-pod#308.

#### Issues closed by this PR

Closes SigNoz/platform-pod#3068

#### Additional Information

We audited `EngineOpts` for more bugs of the same class.
`NoStepSubqueryIntervalFn` is the only field the engine calls without a
nil guard; `promql.NewEngine` defaults the other nil-able fields
(`Parser`, `FeatureRegistry`). The remaining gaps against upstream
wiring are not crashes, and we filed them separately:
SigNoz/pulse-pod#305 (`@` modifier and negative offset disabled),
SigNoz/pulse-pod#306 (engine self-metrics not registered),
SigNoz/pulse-pod#307 (active query tracker startup panic risk),
SigNoz/pulse-pod#309 (step guard in the v3 cache), SigNoz/pulse-pod#310
(upstream proposal to fail fast on the nil callback).

Tests for the bug:

- `pkg/prometheus/engine_test.go` — fails with the exact segfault when
the fix is removed.
- `tests/integration/tests/promqlconformance/04_no_step_subquery.py` — a
step-less subquery through `/api/v5/query_range` returns correct values
on both providers, and the service stays up.
- `tests/integration/tests/alerts/04_promql_subquery_no_step.py` — a
promql alert rule with a step-less subquery evaluates and fires.

🤖 Generated with [Claude Code](https://claude.com/claude-code)
@pull pull Bot locked and limited conversation to collaborators Aug 29, 2026
@pull pull Bot added the ⤵️ pull label Aug 29, 2026
@pull
pull Bot merged commit da9b464 into code:main Aug 29, 2026
2 checks passed
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant