Files
b7a69ebf91 [Test] Add load tests and behavioral checks to incremental upgrade E2E (#4541)
* docs: Add step-by-step process of the basic e2e

Signed-off-by: JiangJiaWei1103 <[email protected]>

* chore: Align test infra setup order with Ray docs

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Add locust load test for incr upgrade e2e

Signed-off-by: JiangJiaWei1103 <[email protected]>

* docs: Improve maintainability of locust yaml

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Remove redundant helper

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Deflake hardcoded sleep for Locust ramp up

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Remove legacy and extract a helper to get rps index

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Ensure remaining traffic routed to the new cluster

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Support high-rps serve application with expected rps over 900

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Test CI

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Test CI

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Test CI

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Recover basic incr upgrade test

Signed-off-by: JiangJiaWei1103 <[email protected]>

* docs: Improve maintainability

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Deflake CI istio gc installation

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Deflake istio gc installation

Signed-off-by: JiangJiaWei1103 <[email protected]>

* revert: Use orig install order

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Support diverse incr upgrade parameter combinations

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Test CI

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Skip transient state check right before promotion

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Retest standard gradual incr upgrade

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Make curl function clearer

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Avoid t.FailNow from non-test goroutines

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Deflake by using commit hash

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Remove redundant checks

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Get rps col index without hardcoded int

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Extract locust warmup constants for tweaking

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Remove redundant line

Signed-off-by: JiangJiaWei1103 <[email protected]>

* test: Split test responsibilities

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Use eg instead of wg

Signed-off-by: JiangJiaWei1103 <[email protected]>

* refactor: Extract trigger incr upgrade helper and curl const

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Deflake waiting for upgrade complete using a longer timeout

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Improve readability

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Remove data race on err btw goroutines

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Fix last migrate time check

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Fix merge and update rollback test for changes in this PR

Signed-off-by: Ryan O'Leary <[email protected]>

* chore: Use a larger runner for higher RPS

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Revert "chore: Use a larger runner for higher RPS"

This reverts commit 067ba1fb3cbf2f7a0cea5e9a2cd129d60779b3dc.

* Revert "[RayService] Rollback Support for Incremental Upgrades (#4109)"

This reverts commit b8b66d42a5.

* codex issue

Signed-off-by: Future-Outlier <[email protected]>

* chore: Deflake by increasing timeout

Signed-off-by: JiangJiaWei1103 <[email protected]>

* chore: Remove worker rsc limits to align with the orig example

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Revert "chore: Remove worker rsc limits to align with the orig example"

This reverts commit 7e821f36dbc922b32dfa23aa171b0e276e29f98f.

* docs: Better strategy naming

Signed-off-by: JiangJiaWei1103 <[email protected]>

* Revert "Revert: Remove rollback e2e"

This reverts commit baf2d6c735ee33e71702b4b9a5a2e70947afabe5, reversing
changes made to 73e66376807d13aaa316be7ac22a7a176f27d285.

* chore: Use Ray project serve config example link

Signed-off-by: JiangJiaWei1103 <[email protected]>

* chore: Deflake CI by lowering steady state RPS

Signed-off-by: JiangJiaWei1103 <[email protected]>

* fix: Make log msg accurate

Co-authored-by: Jun-Hao Wan <[email protected]>
Signed-off-by: Jia-Wei Jiang <[email protected]>

* fix: Make log msg more accurate

Signed-off-by: JiangJiaWei1103 <[email protected]>

* chore: Decrease replicas to align with buildkite runner vcpu

Signed-off-by: JiangJiaWei1103 <[email protected]>

---------

Signed-off-by: JiangJiaWei1103 <[email protected]>
Signed-off-by: Ryan O'Leary <[email protected]>
Signed-off-by: Ryan O'Leary <[email protected]>
Signed-off-by: Future-Outlier <[email protected]>
Signed-off-by: Jia-Wei Jiang <[email protected]>
Co-authored-by: Ryan O'Leary <[email protected]>
Co-authored-by: Ryan O'Leary <[email protected]>
Co-authored-by: Han-Ju Chen (Future-Outlier) <[email protected]>
Co-authored-by: Jun-Hao Wan <[email protected]>
2026-03-21 23:12:40 -07:00
..