perf: pipeline redis set_state writes, cache required-state classes by Alek99 · Pull Request #6745 · reflex-dev/reflex

Alek99 · 2026-07-10T20:49:07Z

All Submissions:

Have you followed the guidelines stated in CONTRIBUTING.md file?
Have you checked to ensure there aren't any other open Pull Requests for the desired changed?

Type of change

Bug fix (non-breaking change which fixes an issue)

Changes To Core Features:

Have you added an explanation of what your changes do and why you'd like us to include them?
Have you written new tests for your core changes, as applicable?

StateManagerRedis.set_state recursed over the substate tree spawning a task per substate; every recursive call re-GETed the same lock key and re-PTTLed it, and each touched state got its own unpipelined SET — roughly 2N+1 redis round trips per event flush for N states. get_state already pipelines its reads; set_state now does the same for writes (ENG-10097):

Verify the lock once, write once: the tree is walked iteratively, the lock GET + PTTL happen a single time, and all touched states' payloads go out in one pipeline. 3 round trips per flush regardless of tree size.
Off-loop pickling: states whose previous payload exceeded 64 KiB are serialized via asyncio.to_thread instead of stalling the event loop (safe: the token's lock is held, so nothing mutates the tree mid-pickle — the old code awaited between pickles too).
Cached _get_required_state_classes: the recursive class-level computation ran on every get_state. It's now an lru_cached module function keyed on (target class, registration context, state hierarchy generation); BaseState.__init_subclass__ and reload_state_module bump the generation, so dynamically created classes and hot reload invalidate cleanly (including same-name class redefinition, where registry sizes don't change).
_was_touched reset after write (mirrors the disk manager's existing get_and_reset_touched_state): previously the flag was never cleared, so a once-touched state kept alive by the oplock cache was re-pickled and re-SET on every subsequent flush. Reset only happens after a successful pipeline execute, so failed writes retry.

Round trips per event flush (modeled on the exact before/after command flows; "3" = lock GET + PTTL + one pipelined write):

state tree	before	after	worst-case serialized latency @ 0.25 ms RTT
root + 3 substates	12	3	3.00 ms → 0.75 ms (4x)
root + 9 substates	30	3	7.50 ms → 0.75 ms (10x)
3 levels × 3 wide (13 states)	39	3	9.75 ms → 0.75 ms (13x)

New unit tests (run against both mock and real redis in CI): test_set_state_verifies_lock_once_and_pipelines_writes counts actual GET/PTTL commands during a 3-substate modify (asserts exactly one of each); test_set_state_resets_touched_flag asserts a second flush of an unchanged instance creates no write pipeline at all; test_set_state_offloads_large_pickles asserts the to_thread path engages above the threshold; test_required_state_classes_cache_invalidation asserts a subclass defined after cache warm-up appears in the cached result via the generation bump. Existing modify_state/oplock/expiration tests cover end-to-end behavior.

Linear: ENG-10097

🤖 Generated with Claude Code

https://claude.ai/code/session_01Sns7EuBPYvfGCxRSRZ1bUx

Generated by Claude Code

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sns7EuBPYvfGCxRSRZ1bUx

greptile-apps · 2026-07-10T20:51:44Z

Greptile Summary

This PR improves performance across state persistence, memo compilation, routing, and hot-path validation. The main changes are:

Redis state writes are batched through a single pipeline.
Required Redis state-class lookup is cached with hierarchy invalidation.
Memo component generation reuses duplicate passthrough definitions.
Computed-var validation and dirty propagation skip repeated hot-path work.
Mutable proxy instances and websocket client metadata are cached.

Confidence Score: 5/5

This looks safe to merge.

No blocking issues found in the changed code.

Important Files Changed

Filename	Overview
reflex/istate/manager/redis.py	Redis persistence now collects touched states, serializes them, writes them through one pipeline, and clears touched flags after a successful write.
packages/reflex-base/src/reflex_base/event/processor/base_state_processor.py	Final chained-event routing can now run after the state lock is released while deltas still emit inline.
reflex/istate/proxy.py	Mutable proxy reads now reuse cached proxies and skip wrapping immutable values.

_{Reviews (2): Last reviewed commit: "Appease pyright in redis manager tests" | Re-trigger Greptile}

codspeed-hq · 2026-07-10T20:52:57Z

Merging this PR will not alter performance

✅ 27 untouched benchmarks
⏩ 8 skipped benchmarks¹

_{Comparing claude/reflex-perf-optimizations-21kg8y-redis (4c6cc4f) with claude/reflex-perf-optimizations-21kg8y-socket (7ef34b1)}

8 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9cd59d2f22

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

masenf · 2026-07-13T22:41:58Z

+            # Reset the touched flag so an unchanged state is not re-pickled
+            # and re-written by subsequent flushes of the same instance.
+            for substate in written_states:
+                substate._was_touched = False


the _was_touched flag was already being reset by virtue of BaseState.__setstate__ popping the flag out of the payload that gets saved in the pickle, so this is just added overhead now.

set_state recursed over the substate tree spawning a task per substate; every recursive call re-GETed the same lock key and re-PTTLed it, and each touched state got its own unpipelined SET -- roughly 2N+1 round trips per event flush for N states. The tree is now walked once, the lock is verified once, and all touched states are written in a single pipeline: 3 round trips per flush regardless of tree size (12->3 for a root with 3 substates, 39->3 for a 13-state tree). Also: - Pickling runs off the event loop (asyncio.to_thread) for states whose previous payload exceeded 64KiB, instead of stalling the loop. - The recursive required-state-classes computation (pure class-level data) is now lru-cached per registration context and state hierarchy generation; subclass creation and module hot reload bump the generation, so dynamic classes invalidate cleanly. - _was_touched is reset after a successful write (mirroring the disk manager), so a once-touched state held by the oplock cache is no longer re-pickled and re-SET on every subsequent flush. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sns7EuBPYvfGCxRSRZ1bUx

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sns7EuBPYvfGCxRSRZ1bUx

…mock getdel

Alek99 · 2026-07-18T01:53:07Z

Root cause of the unit-tests failures (4× test_app_modify_state_clean, all Python versions):

The unit-tests job exports REFLEX_REDIS_URL, so StateManager.create() returns StateManagerRedis. This test's assert sub._was_touched / assert root_state._was_touched (after the modify_state context exits) now fail because the pipelined set_state in this PR resets _was_touched = False after a successful write — which is the intended fix (so an oplock-cached state isn't re-pickled/re-SET on every flush), and matches the disk manager's existing get_and_reset_touched_state.

So _was_touched no longer survives persistence for the redis manager. The _was_touched assertions in test_app_modify_state_clean need to reflect that — e.g. capture the flag inside the async with block before the write, or assert on the emitted delta (which the frontend branch already does) rather than the post-write flag. Flagging rather than pushing since the branch was force-pushed into the consolidated stack and this same test file was just touched in 66c2c99f; happy to push a targeted fix if you'd like me to.

Generated by Claude Code

Alek99 requested a review from a team as a code owner July 10, 2026 20:49

Alek99 pushed a commit that referenced this pull request Jul 10, 2026

chore: add news fragment for #6745

c4b7b46

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sns7EuBPYvfGCxRSRZ1bUx

greptile-apps Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread reflex/istate/manager/redis.py

chatgpt-codex-connector Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread reflex/istate/manager/redis.py

masenf reviewed Jul 13, 2026

View reviewed changes

claude and others added 5 commits July 17, 2026 18:44

chore: add news fragment for #6745

d95422b

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sns7EuBPYvfGCxRSRZ1bUx

Expect a single held-too-long warning now that set_state is pipelined

591f7b8

Fix pipelined set_state test: distinct substate vars, self-contained …

66c2c99

…mock getdel

Appease pyright in redis manager tests

4c6cc4f

Alek99 force-pushed the claude/reflex-perf-optimizations-21kg8y-redis branch from c4b7b46 to 4c6cc4f Compare July 18, 2026 01:46

Alek99 changed the base branch from main to claude/reflex-perf-optimizations-21kg8y-socket July 18, 2026 01:46

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

perf: pipeline redis set_state writes, cache required-state classes#6745

perf: pipeline redis set_state writes, cache required-state classes#6745
Alek99 wants to merge 5 commits into
claude/reflex-perf-optimizations-21kg8y-socketfrom
claude/reflex-perf-optimizations-21kg8y-redis

Alek99 commented Jul 10, 2026

Uh oh!

greptile-apps Bot commented Jul 10, 2026 •

edited

Loading

Uh oh!

Uh oh!

codspeed-hq Bot commented Jul 10, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

masenf Jul 13, 2026

Uh oh!

Alek99 commented Jul 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

Alek99 commented Jul 10, 2026

All Submissions:

Type of change

Changes To Core Features:

Uh oh!

greptile-apps Bot commented Jul 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Uh oh!

Uh oh!

codspeed-hq Bot commented Jul 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merging this PR will not alter performance

Footnotes

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

masenf Jul 13, 2026

Choose a reason for hiding this comment

Uh oh!

Alek99 commented Jul 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

greptile-apps Bot commented Jul 10, 2026 •

edited

Loading

codspeed-hq Bot commented Jul 10, 2026 •

edited

Loading