[Streams] Add scalability performance journeys for Streams

[rStelmach]

Summary

Adds six @kbn/journeys performance journeys that validate the Streams feature at scale, covering the primary user flows across listing, detail, and management pages. These journeys are excluded from PR CI and run only in scheduled performance pipelines. The heavy wired-hierarchy journey (streams_wired_hierarchy) runs in a dedicated weekly Streams-only pipeline due to its large data setup.

Journeys

Journey	Scale	What it exercises
streams_listing_page	5 000 classic + 3 wired children	Load, search, expand/collapse, navigate to detail
streams_data_quality	3 wired children	Navigate to data quality tab, verify KPI metrics
streams_processing_step	3 wired children	Open processor form, configure grok processor, save
streams_retention	3 wired children	Open retention modal, toggle inherit, set custom retention
streams_field_mapping	3 wired children + 200 fields	Open schema flyout, add keyword field, review & submit
streams_wired_hierarchy	1 000 wired children	Expand/collapse large tree, search children, navigate to detail

Data creation strategies

• Classic streams at scale: Uses ES _bulk API to auto-create 5 000 unmanaged data streams in batches of 250, bypassing the Streams backend global lock. Raises cluster.max_shards_per_node to accommodate the shard count.
• Wired hierarchy at scale: Uses batched content pack imports (20 batches of 50 children) with retry logic for transient 409/422 errors, followed by a final root routing update via the ingest API. Handles idempotency for "already exists" conflicts caused by timed-out-but-successful prior attempts.
• Small wired hierarchy: Serial fork of 3 children from logs.otel for the lightweight journeys.

Infrastructure

• .buildkite/pipelines/performance/streams_weekly.yml — weekly Streams-only pipeline (JOURNEYS_GROUP=streams) on kb-static-scalability-2
• .buildkite/pipeline-resource-definitions/kibana-streams-performance-weekly.yml + .buildkite/pipeline-resource-definitions/locations.yml — Buildkite pipeline resource definition
• streams_heavy_config.ts — extended FTR config with 1-hour mocha timeout (covers beforeSteps data setup)
• streams journey group added to run_performance_cli.ts for --group streams execution
• Metrics are reported via report_performance_metrics.sh.

Follow-ups

• Add a dedicated classic-stream mapping-at-scale journey that exercises very large field counts (up to 10k), in a separate PR.

Summary by CodeRabbit

• Tests

  • Added comprehensive Streams performance journeys (listing, data quality, field mapping, processing, retention, wired hierarchy), supporting heavy-profile runs with extended timeouts and a new journey group; included extensive setup utilities for bulk data, large wired hierarchies, and scaled test orchestration.

• Chores

  • Updated test metadata and scheduling to disable certain streams jobs in scheduled pipelines, expanded project references/dependencies, and added ownership entries for Streams performance tests.

Buildkite run

https://buildkite.com/elastic/kibana-single-user-performance/builds/19020#019d002d-9541-4221-b66d-ad31c1b71df0

[delanni]

I'm not sure if this should be true. Afaik, this means something like: "start a build when a new branch appears". Let's start with this and observe if it starts builds for random branches

[rStelmach]

Good catch, changed to false. It’s intended to run scheduled weekly on main, no need to have true here

[elasticmachine]

💛 Build succeeded, but was flaky

• Buildkite Build
• Commit: 442a790

Failed CI Steps

• Jest Tests #5

Test Failures

• [job] [logs] Jest Tests #5 / form payload & api errors should surface API errors and send the correct payload on success

Metrics [docs]

Unknown metric groups

ESLint disabled line counts

id	before	after	diff
@kbn/test-suites-xpack-performance	4	5	+1

Total ESLint disabled count

id	before	after	diff
@kbn/test-suites-xpack-performance	4	5	+1

History

• 💔 Build #414592 failed cd8fd2e
• 💔 Build #414370 failed b3db365
• 💛 Build #413524 was flaky 599ec21
• 💔 Build #413092 failed b21a7ef
• 💔 Build #410747 failed 5e4d806
• 💔 Build #410721 failed fd1a689

[flash1293]

Seems like the description isn't fully up to date anymore.

About the mapping - for classic streams we should test with more fields (up to 10k), it's something that does happen in practice (but can also happen on a separate PR).

What's our story for getting notified about these / acting on them?

[flash1293]

Looks like this isn't used anywhere?

[rStelmach]

Slipped through during cleanup, thanks for catching this

[rStelmach]

@flash1293

I think it's best to handle that in a separate PR, once this one is merged, we'll have the dedicated pipeline set up, which will make it easier to test on the exact environment these journeys will run in. I'll create a follow-up PR after this lands.

What's our story for getting notified about these / acting on them?

There's a separate issue for that: https://github.com/elastic/streams-program/issues/938. Once this PR is merged and we have some data flowing, I'll start working on this

[macroscopeapp]

🟡 Medium synthtrace_data/streams_data.ts:333

ensureScaleParentStream calls forkStream without retry logic for HTTP 422 lock contention. When the Streams backend is under load, the fork can fail immediately and propagate the error, causing setupLargeWiredHierarchy to fail even though other mutation operations in this file implement exponential backoff retries for the same condition.

🤖 Copy this AI Prompt to have your agent fix this:

In file x-pack/performance/synthtrace_data/streams_data.ts around line 333:

`ensureScaleParentStream` calls `forkStream` without retry logic for HTTP 422 lock contention. When the Streams backend is under load, the fork can fail immediately and propagate the error, causing `setupLargeWiredHierarchy` to fail even though other mutation operations in this file implement exponential backoff retries for the same condition.

Evidence trail:
x-pack/performance/synthtrace_data/streams_data.ts lines 333-350 (ensureScaleParentStream with no retry); lines 127-161 (createSingleClassicStream with retry logic for isLockContentionError); lines 398-410 (fork loop with lock contention retry); lines 786-806 (setupLargeWiredHierarchy calling ensureScaleParentStream)

[macroscopeapp]

Approvability

Verdict: Needs human review

1 blocking correctness issue found. CODEOWNERS file was modified by a non-owner — requires human review

^{You can customize Macroscope's approvability policy. Learn more.}