docs: stream replay caps + get_transaction profiling (Hyperion 4.1.0) - #11
Merged
Conversation
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
Contributor
There was a problem hiding this comment.
Pull request overview
Documents remaining Hyperion v4.1.0 operator/client behaviors around stream history replays and get_transaction profiling, extending the existing chain configuration reference and the stream client guide.
Changes:
- Adds new API config reference entries for
hot_first_transaction_profiling,stream_scroll_limit, andstream_max_concurrent_replays, including defaults and operational rationale. - Updates the example
config.jsonsnippet to include the new options. - Adds a stream client warning that
start_fromhistory replays are subject to server-side document and concurrency limits, with client retry guidance.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| docs/providers/setup/chain.md | Documents new v4.1.0 API config options and adds them to the config example. |
| docs/dev/stream_client.md | Adds a client-facing warning about server-side replay limits and suggested client behavior. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| - `"hot_first_transaction": false` ⇒ When enabled, a `get_transaction` lookup **without** a `block_hint` searches the most recent action partition(s) first and widens to the full index set **only** on a miss. Without it, a `trx_id` lookup has no block range to prune on and fans out across every action partition — including cold/old shards. Because all of a transaction's documents share one block (one partition), a hot-window hit is always complete, so the common case (recent transactions) never touches the older tiers. Reuses `hot_first_window`. Default off; safe to enable without reindexing. *(Tip: clients that already know the block can pass `?block_hint=<block_num>` to target the single partition directly.)* | ||
| - `"hot_first_transaction_profiling": false` ⇒ Diagnostic for sizing `hot_first_window` from real traffic *(Hyperion ≥ 4.1.0)*. When enabled, every `get_transaction` served **without** a `block_hint` logs one `[gtx-profile]` line: which phase answered it (`hot` or `full`), per-phase Elasticsearch timings, and `parts_back` — how many partitions older than head the transaction was found in (a window of `parts_back + 1` would have served it from the hot path; `-1` = not found). Aggregate `parts_back` over a sample to see the age distribution of lookups. One log line **per request** — enable briefly to sample, then disable. | ||
| - `"stream_scroll_limit": 50000` ⇒ Maximum documents a single streaming **history replay** (a subscription with `start_from` in the past) may scroll *(Hyperion ≥ 4.1.0; previously unlimited when unset)*. A replay walks every index partition — including cold-tier shards on tiered clusters — so an unbounded deep replay can saturate old storage. Set `-1` for unlimited (a warning is logged past 100k documents per replay). | ||
| - `"stream_max_concurrent_replays": 4` ⇒ Maximum streaming history replays running concurrently per API process *(Hyperion ≥ 4.1.0)*. Additional replay requests are rejected with a "server busy, please retry" message instead of piling scroll load onto the cluster (e.g. during client reconnect storms). Applies regardless of `stream_scroll_limit`. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Documents the remaining Hyperion v4.1.0 config options not yet covered:
api.stream_scroll_limit— now defaults to 50000 per history replay when unset (previously unlimited);-1keeps unlimited with a warning past 100k docs (hyperion-history-api #178).api.stream_max_concurrent_replays— per-process concurrency cap (default 4) with "server busy, retry" rejection (#178).api.hot_first_transaction_profiling— opt-in[gtx-profile]diagnostic for sizinghot_first_windowfrom measuredparts_backdistribution (#181).Also adds a client-facing note in the stream client docs:
start_fromreplays are subject to these server-side limits; retry with backoff on "server busy", and prefer bounded REST pagination for deep history reads.With this, all operator-facing v4.1.0 options are documented (hot-first #7/#8, delta deletions #9, asc bounds #10 already merged).