Fix activities missing when sleep dbt build times out

[Asherlc]

Summary

Production /activities was stale because sleep_heart_rate_sample microbatch runs timed out (121 daily batches with a 120-day lookback), failing the entire dbt build and blocking deduped_activities from refreshing. This splits analytics-worker dbt into activity and sleep/dashboard phases so activity read models keep updating even when sleep builds fail. It also rewrites sleep_heart_rate_sample from microbatch to dirty-key incremental updates, scoping work to changed sleeps, sensors, and overlapping activities instead of reprocessing the full lookback every run. Tests were updated for the split build and dirty-key sleep model pattern.

Test plan

• Verify read_model_microbatch.sql.test.ts passes locally
• After deploy, confirm deduped_activities refreshes even if sleep models fail
• Confirm missing activity 550063f5-bcd1-4c63-88bf-c4533b62b9c5 appears on /activities
• Monitor first sleep_heart_rate_sample run after deploy (120-day bootstrap may be heavy, then incremental runs should be fast)

Made with Cursor

---

Summary by cubic

Splits analytics builds into activity and sleep/dashboard phases and rewrites sleep_heart_rate_sample to an append-based dirty-key incremental model with ReplacingMergeTree(refresh_version). This avoids microbatch timeouts, keeps /activities fresh, and fixes the table’s ClickHouse sort/order stability.

• Bug Fixes

  • Activity read models keep refreshing even if sleep builds fail by running the activity phase first.
  • Stabilized timestamps in sleep_heart_rate_sample using a refresh_clock and source_refreshed_at to avoid nondeterminism.
  • Ensured non-null key fields with assumeNotNull for sleep_id, user_id, recorded_at, and recorded_date.
  • Fixed activity visibility unit guard to prevent incorrect hiding/showing of activities.
  • Corrected ClickHouse sort key for sleep_heart_rate_sample to improve merges and determinism.

• Refactors

  • Split analytics-worker builds in entrypoint.sh into DBT_ACTIVITY_MODELS and DBT_SLEEP_DASHBOARD_MODELS, combined into DBT_SAFE_MODELS and executed via run_dbt_safe_builds.
  • Replaced sleep_heart_rate_sample microbatch with append + ReplacingMergeTree(refresh_version) and dirty-key increments from changed sleeps (_peerdb_synced_at), heart rate updates (refreshed_at), overlapping activities, and deleted/absent records; emit tombstones for stale samples and include an initial_lookback_days bootstrap.

^{Written for commit 3a38c18. Summary will update on new commits.}

Summary by CodeRabbit

• Bug Fixes

  • Improved data freshness and consistency for sleep-related metrics with updated incremental processing logic and stale data handling.

• Chores

  • Updated backend orchestration to optimize build sequencing for improved performance and reliability.

• Tests

  • Enhanced test coverage to validate incremental data processing and model behavior.

[qodo-code-review]

Qodo reviews are paused for this user.

Troubleshooting steps vary by plan Learn more →

On a Teams plan?
Reviews resume once this user has a paid seat and their Git account is linked in Qodo.
Link Git account →

Using GitHub Enterprise Server, GitLab Self-Managed, or Bitbucket Data Center?
These require an Enterprise plan - Contact us
Contact us →

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[sourcery-ai]

Sorry @Asherlc, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 7b01db5e-a433-4434-9173-8ad3702cd8ef

📥 Commits

Reviewing files that changed from the base of the PR and between 313fd36 and 3a38c18.

📒 Files selected for processing (3)

• analytics/models/read_models/read_model_microbatch.sql.test.ts
• analytics/models/read_models/sleep_heart_rate_sample.sql
• entrypoint.sh

---

📝 Walkthrough

Walkthrough

sleep_heart_rate_sample.sql switches from microbatch to append incremental strategy, replacing the prior simple CTE with a dirty-key pipeline that computes affected (user_id, sleep_id) pairs from sleep/sensor/activity changes and emits both current rows and tombstone rows (is_deleted=1). entrypoint.sh adds run_dbt_safe_builds(), splitting safe builds into a sequential activity-then-sleep/dashboard pair. Tests are updated to match both changes.

Changes

sleep_heart_rate_sample rewrite and entrypoint build split

Layer / File(s)	Summary
entrypoint.sh: split safe builds into activity-then-sleep/dashboard entrypoint.sh, analytics/models/read_models/read_model_microbatch.sql.test.ts	Replaces the single DBT_SAFE_MODELS selector with DBT_ACTIVITY_MODELS and DBT_SLEEP_DASHBOARD_MODELS env vars; introduces run_dbt_safe_builds() that runs two sequential dbt build commands; updates sync, analytics, and analytics-worker modes; test asserts ordering, combined selector, and --select args.
sleep_heart_rate_sample: switch to append + incremental state setup analytics/models/read_models/sleep_heart_rate_sample.sql	incremental_strategy changes from microbatch to append, microbatch params removed; initial_lookback_days variable and target_state CTE added to compute last_refreshed_at/is_empty for gating downstream filters.
sleep_heart_rate_sample: dirty-key pipeline analytics/models/read_models/sleep_heart_rate_sample.sql	Defines sleep_source, active_sleep, sleep_dirty_keys, sensor_dirty_keys (from changed heart-rate samples), activity_dirty_keys, initial_dirty_keys, and stale_sleep_dirty_keys (left-join to detect sessions needing tombstones).
sleep_heart_rate_sample: current samples, tombstones, and final output analytics/models/read_models/sleep_heart_rate_sample.sql, analytics/models/read_models/read_model_microbatch.sql.test.ts	Adds sleep_activity_bounds/active_activity exclusion logic, current_samples CTE, existing_samples/stale_sample_tombstones for deletion rows, refresh_clock, and final UNION ALL of is_deleted=0 current rows and is_deleted=1 tombstones; tests assert the append strategy, dirty-key derivation, tombstone presence, and absence of microbatch/lookback patterns.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

Possibly related PRs

• Asherlc/dofek#1185: Introduced the microbatch design in sleep_heart_rate_sample.sql that this PR replaces with the append/dirty-key approach.
• Asherlc/dofek#1183: Modifies entrypoint.sh's dbt safe-model selection, directly overlapping with the DBT_ACTIVITY_MODELS/DBT_SLEEP_DASHBOARD_MODELS split made here.
• Asherlc/dofek#1190: Changes entrypoint.sh and the read-model microbatch test suite around how dbt safe builds execute per worker, intersecting with run_dbt_safe_builds() ordering added here.

Suggested labels

area/infra, type/bug

Suggested reviewers

• cubic-dev-ai

🚥 Pre-merge checks | ✅ 2

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title directly addresses the main issue (activities missing due to sleep dbt build timeouts) and reflects the core solution (splitting builds to isolate failures).

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

Storybook previews for a0f81a3b are ready:

• Web Storybook
• Mobile Storybook

This comment updates automatically on each PR push.

[cubic-dev-ai]

3 issues found across 3 files

Confidence score: 2/5

• In entrypoint.sh, the analytics-worker dbt phase commands are not chained, so an early phase failure can be masked and the cycle may report success despite failed activity models; this undermines fail-fast behavior and can hide broken runs in production — chain the phase commands (or explicitly propagate exit codes) before merging.
• In analytics/models/read_models/sleep_heart_rate_sample.sql, dirty-key logic misses activity removals (deleted/provider-absent transitions), so overlapping sleep windows are not recomputed and stale sleep heart-rate rows can persist until unrelated updates occur — include removal transitions in dirty-key detection before merge.
• In analytics/models/read_models/sleep_heart_rate_sample.sql, the incremental watermark uses refreshed_at without activity _peerdb_synced_at, so activity-only changes can stall progress and trigger repeated reprocessing of the same sleep keys each run — incorporate the activity sync timestamp into watermark advancement to de-risk merge.

_{You’re at about 94% of the monthly reviewed-line limit. You may want to disable incremental reviews to conserve quota. Reviews will continue until that limit is exceeded. If you need help avoiding interruptions, please contact contact@cubic.dev.}

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

_{You’re at about 95% of the monthly reviewed-line limit. You may want to disable incremental reviews to conserve quota. Reviews will continue until that limit is exceeded. If you need help avoiding interruptions, please contact contact@cubic.dev.}

<sub>Requires human review: Auto-approval blocked by 3 unresolved issues from previous reviews.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

_{You’re at about 95% of the monthly reviewed-line limit. You may want to disable incremental reviews to conserve quota. Reviews will continue until that limit is exceeded. If you need help avoiding interruptions, please contact contact@cubic.dev.}

<sub>Requires human review: Auto-approval blocked by 2 unresolved issues from previous reviews.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 2 files (changes from recent commits).

_{You’re at about 95% of the monthly reviewed-line limit. You may want to disable incremental reviews to conserve quota. Reviews will continue until that limit is exceeded. If you need help avoiding interruptions, please contact contact@cubic.dev.}

<sub>Requires human review: Auto-approval blocked by 2 unresolved issues from previous reviews.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 3 files (changes from recent commits).

_{You’re at about 95% of the monthly reviewed-line limit. You may want to disable incremental reviews to conserve quota. Reviews will continue until that limit is exceeded. If you need help avoiding interruptions, please contact contact@cubic.dev.}

<sub>Requires human review: Restructures dbt orchestration, updates sleep_heart_rate_sample to an incremental pattern, and changes production analytics pipelines; may impact data freshness and metrics, so human review is required.

Re-trigger cubic</sub>