26.4.1

New Features ✨

• Add default value to attribute aggregations by @wmak in #7888
• Make BLQ stale threshold and static friction runtime-configurable by @kylemumma in #7880

Bug Fixes 🐛

• (admin) Allow ClickHouse replace* functions in read-only query validation by @xurui-c in #7876
• (COGS) Unique labels for resource_ids by @xurui-c in #7864
• (outcomes) Deduplicate outcomes by item_id within batch by @untitaker in #7885

Internal Changes 🔧

Eap Outcomes

• Add min_partition tag by @MeredithAnya in #7891
• Add duplicate_item_count metric by @MeredithAnya in #7887

Other

• (ci) Pull out migrations tests maybe lets see by @MeredithAnya in #7881
• (gocd) Bump gocd-jsonnet to v3.0.0 (grouped pipedream) by @mchen-sentry in #7878
• Bump new development version by @sentry-release-bot[bot] in 41fe9b16

26.4.0

New Features ✨

Consumer

• Emit batch_write_bytes metric from RowBinary writer by @onewland in #7863
• Add memory-based batching limit to Rust consumer by @onewland in #7854

Other

• (cross-item) Prepare settings for production by @volokluev in #7706
• (docker) Switch distroless images to DHI by @oioki in #7821
• (manual-jobs) Add job to rerun completed idempotent migrations by @onewland in #7859
• (observability) Add baseline tags to RPC sourced sentry errors by @volokluev in #7845
• (options) Sentry options scaffolding by @kenzoengineer in #7836
• (outcomes) Add no_parent_span outcome by @Lms24 in #7866
• (snql) Add Sentry error tags for SnQL API requests by @volokluev in #7839
• (timeseries) Add 6h interval by @wmak in #7867
• Blq feature flag can be updated without restart by @kylemumma in #7871
• Add a backlog-queue to the snuba consumers by @kylemumma in #7856
• Debugger support for consumers, local metrics support, bugbear by @kylemumma in #7818

Bug Fixes 🐛

Rpc

• Honor deprecated attribute keys in trace item attribute values by @xurui-c in #7853
• Validate that TraceItemTable requests have at least one column by @volokluev in #7837

Other

• (ci) Bump timeout and use --no-project on bump_version test by @kenzoengineer in #7841
• (COGS) Label metrics correctly for query side by @xurui-c in #7826
• (slo) Correctly report TOO_MANY_BYTES from allocation policy as rate-limited by @volokluev in #7846
• (snuba commit log) Modify Rust consumer to produce next message offset to Snuba commit log by @shashjar in #7852

Internal Changes 🔧

Eap Outcomes

• Use sentry-options to change which timestamp to use by @kenzoengineer in #7838
• Add category_metrics by @MeredithAnya in #7847
• Add config to change which timestamp to use by @MeredithAnya in #7834
• Debug -> info by @MeredithAnya in #7824

Other

• (admin) Add skip_on_cluster and cluster_name option for target host by @MeredithAnya in #7823
• (ci) Add cancel-in-progress to pull_request prs by @joshuarli in #7873
• (deps) Bump picomatch from 2.3.1 to 2.3.2 in /snuba/admin by @dependabot in #7840
• (eap) Add arrayExists filter for attribute name queries by @volokluev in #7832
• (INC-2103) Increase distributed_ddl_task_timeout to 300 by @MeredithAnya in #7862
• (metrics) Bump sentry-relay version 0.9.25 by @k-fish in #7825
• Bump sentry-arroyo to 2.38.7 by @getsentry-bot in #7857
• Bump sentry-arroyo to 2.38.5 by @getsentry-bot in #7851
• Bump rust-toolchain to 1.88.0 by @MeredithAnya in #7835

Other

• Volo/fix bools autocomplete by @volokluev in #7848

26.3.1

ref(eap-outcomes): updating committing logic, more metrics (#7813)

## Summary

This PR improves the committing and backpressure handling in the
accepted outcomes consumer, and adds a `--commit-frequency-sec`
parameter to control how often commit requests are emitted.

### Changes

- **`CommitOutcomes`**: Added a configurable `commit_frequency` (default
10s) so that offsets are only committed on a timer rather than on every
poll. Offsets are now only tracked after a successful produce — if the
produce step rejects a message, offsets are **not** advanced. On `join`,
a forced commit is issued regardless of the timer.

- **`OutcomesAggregator`**: Fixed backpressure handling. Previously,
when the next step rejected a message during flush, the batch was
silently restored and the flush effectively became a no-op. Now, a
rejected message is carried over and retried on the next `poll`. While a
message is carried over, incoming `submit` calls return
`MessageRejected` so upstream applies backpressure. The `join` path also
retries carried-over messages until they succeed or the deadline
elapses.

- **`ProduceOutcome`**: Added a `timer!` metric
(`accepted_outcomes.batch_produce_ms`) to track how long each batch
produce takes.

- **`accepted_outcomes_consumer` CLI**: Added `--commit-frequency-sec`
(default `10`) that is plumbed through to `CommitOutcomes`.

### Future considerations

- Might have this just run `RunTask` instead of `RunTaskInThreads` given
that producing should be pretty quick and batching will be what takes
longer, might be better to prevent flushing a batch until the first once
is done producing. If we have multiple threads and something fails that
would be more data loss if we are committing early

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

26.3.0

fix(cogs): label metrics correctly (#7815)

26.2.1

ref(tests): refactor migration validator tests (#7754)

As a continuation of my journey to see where we are spending our time in
tests, it seems like a lot of time is spent in the migrations tests, or
really any test that requires `clickhouse_db` marker. Some other PRs aim
to run only whats neccesary, see
https://github.com/getsentry/snuba/pull/7749 and
https://github.com/getsentry/snuba/pull/7751.

Here `test_validate_all_migrations` is iterating over all the migrations
to validate the order, and the only reason it was using the
`clickhouse_db` marker is because it was querying clickhouse in order to
do some logic to get the local name for a dist table. We don't need to
do all that, we can do what we do in admin
https://github.com/getsentry/snuba/blob/5fd5b75ea00fd7c30f631aedf5b33598045d267a/snuba/admin/clickhouse/nodes.py#L31
by looking at the table schema for the storages associated with the
storage set key

26.2.0

instrument gettrace endpoint more (#7741)

Approach:
- Goal: get a high-level signal on where the slowdown is (e.g., validate
whether post-processing is the issue), not detailed diagnostics (what is
the longest it took to process a row or how many attributes such a row
has)
- Plan to iterate across multiple PRs to progressively narrow down the
bottleneck.
- Avoid instrumenting every part of the endpoint to prevent trace
waterfall bloat.
- Hypothesis: _process_results is the likely bottleneck, so
instrumentation is focused there.
- Instrumentation is intentionally lightweight until we have stronger
evidence.
- Some instrumentation may be removed as we rule out potential
bottlenecks.

<img width="792" height="643" alt="Screenshot 2026-02-15 at 5 17 26 PM"
src="https://github.com/user-attachments/assets/8ea1a784-3867-4846-bec0-06cff5a04419"
/>

---------

Co-authored-by: Rachel Chen <rachelchen@MacBookPro.attlocal.net>
Co-authored-by: Rachel Chen <rachelchen@PL6VFX9HP4.local>

26.1.0

New Features ✨

• (ast) Add ArbitrarySQL AST node to allow for subquery optimization by @volokluev in #7636
• (capman) Add detailed observability to allocation policy span data by @volokluev in #7638
• (consumers) Send logs for tracing events at or above INFO by @lcian in #7250
• (eap) Gdpr export endpoint by @xurui-c in #7586
• (health) Allow clickhouse check bypass by @onewland in #7600
• (outcomes) Add 'ignored' reason to client discard reasons list by @Lms24 in #7635
• Add metric that tracks "partitions left to optimize" by @kylemumma in #7620

Build / dependencies / internal 🔧

Deps

• Update rdkafka to 0.38 and librdkafka to 2.10.0 by @Dav1dde in #7630
• Bump docker/setup-buildx-action from 3.11.1 to 3.12.0 by @dependabot in #7608
• Bump js-yaml from 3.14.1 to 3.14.2 in /snuba/admin by @dependabot in #7544
• Bump tracing-subscriber from 0.3.18 to 0.3.20 in /rust_snuba by @dependabot in #7613
• Bump actions/upload-artifact from 5 to 6 by @dependabot in #7593
• Bump actions/download-artifact from 6 to 7 by @dependabot in #7594
• Bump actions/cache from 4 to 5 by @dependabot in #7595

Other

• (deletes) Make error log a metric instead by @MeredithAnya in #7627
• (gocd) Bump gocd lib version to v2.18.0 by @dmajere in #7617
• (metrics) Don't write the histogram timing metrics by @volokluev in #7616
• Upgrade to action-setup-venv 3.2.0 by @joshuarli in #7607

Other

• imp(deletes): ongoing mutations check by @onewland in #7639
• prodeng-605: migrate uwsgi to granian by @gi0baro in #7566
• Kill Veneur metrics for good by @volokluev in #7637
• remove documentation from the rust test image by @volokluev in #7633
• upgrade datadog client by @volokluev in #7554
• debug(deletes): log out data so we can inspect allowlist not working by @onewland in #7626
• cleanup(ci): remove useless test and useless function by @volokluev in #7618
• cleanup(gocd): remove pipeline that deploys everything by @onewland in #7587
• deploy: add eap-items LW delete consumer to deploy list in non-STs by @onewland in #7612
• meta: Bump new development version in ddc2e223

25.12.1

New Features ✨

• feat(deletes): add EAP deletes allowlist for launch by @onewland in #7591

• feat(eap): record bytes ingested by item type by @onewland in #7604

Bug Fixes 🐛

• fix notHandled errors for search issues and transactions by @volokluev in #7599

Build / dependencies / internal 🔧

• chore(eap): Try to understand the number of trace IDs we're fetching by @phacops in #7603

• chore(snuba): take new_datadog out of metrics config by @volokluev in #7598

Other

• ref(admin): allow system queries on nodes sans clusters by @MeredithAnya in #7605
• Update distribution metric naming convention by @volokluev in #7602
• cleanup: emit item type in metrics as string by @onewland in #7588
• cleanup: remove stale docker-compose by @onewland in #7597
• meta: Bump new development version in 8e2624b3
• ref(eap): kill snuba.web.rpc.v1.resolvers.R_eap_items.common.common by @onewland in #7590

25.12.0

New Features ✨

• feat(deletes): support item attribute conditions in API and consumer by @onewland in #7570

Eap

• feat(eap): Allow arrays to be queried from EAP by @phacops in #7551

• feat(eap): emit item type counts by @onewland in #7548

• feat(oncall): don't send transient cache set errors to sentry, emit metric by @volokluev in #7560

• feat(self-hosted): Manually run post release script by @hubertdeng123 in #7563

• feat: support attribute allowlist in traceitemstats by @kylemumma in #7568

Bug Fixes 🐛

• fix(image): Ensure we install dev dependencies by @hubertdeng123 in #7578

• fix: Remove "performance" code owners by @gggritso in #7584

Build / dependencies / internal 🔧

Deps

• chore(deps): bump actions/create-github-app-token from 2.1.4 to 2.2.0 by @dependabot in #7550

• chore(deps): bump actions/dependency-review-action from 4.8.1 to 4.8.2 by @dependabot in #7539

• chore(deps): bump actions/checkout from 5 to 6 by @dependabot in #7549

• chore(gocd): remove outcomes-billing consumer from python deploys by @onewland in #7564

• ci: leaner snuba image by @aldy505 in #7542

Other

• ref(eap): fix outcomes based routing retention downsampling by @MeredithAnya in #7592
• ref: bump sentry-arroyo to 2.35.0 by @getsentry-bot in #7583
• Change new_datadog timing to send distribution metrics by @volokluev in #7582
• Double write metrics to new datadog by @volokluev in #7555
• ref: remove build_target for GAR image by @aldy505 in #7571
• ref(redis): increase changes limit from 25 to 100 by @MeredithAnya in #7572
• ref: bump sentry-protos to 0.4.8 by @getsentry-bot in #7569
• ref: bump sentry-arroyo to 2.34.0 by @getsentry-bot in #7567
• dep: bump granian to 2.6 by @gi0baro in #7565

25.11.1

Build / dependencies / internal 🔧

• ref: bump sentry-protos to 0.4.7 by @getsentry-bot in #7559
• chore: enable trace pagination support by default by @kylemumma in #7558
• ref: bump sentry-protos to 0.4.6 by @getsentry-bot in #7553
• ref(consumer): change BytesInsertBatch to support builder pattern by @onewland in #7479

New Features ✨

• feat(eap): record item type counts in batches by @onewland in #7481
• feat(processor): Handle casted columns in UUID column processor by @Zylphrex in #7552

Bug Fixes 🐛

• fix: subscription executors should not crash when there are too many concurrent queries by @xurui-c in #7547
• fix(gocd): fix gocd filters for rs/py deploys by @onewland in #7545

Other

• ref(runtime-config): update write_node_replacements_global default by @MeredithAnya in #7556
• feat(query-pipeline): Add configs to customize maximum size of a query by @phacops in #7546
• ref(runtime-config): remove *_matview_version by @MeredithAnya in #7543
• ref(runtime-config): remove use_split by @MeredithAnya in #7540
• meta: Bump new development version in 2fdaba59
• ref(runtime-config): remove use_readthrough_query_cache by @MeredithAnya in #7541