Make warmup settings dynamic for pull-based ingestion by kaustubhbutte17 · Pull Request #20936 · opensearch-project/OpenSearch

kaustubhbutte17 · 2026-03-20T03:07:28Z

Description

Follow-up to #20526. Changes the two warmup settings from Final (immutable after index creation) to Dynamic (updatable at runtime via the update settings API).

This allows operators to adjust warmup behavior without recreating the index.

Settings Changed

Setting	Before	After
`index.ingestion_source.warmup.timeout`	Final	Dynamic
`index.ingestion_source.warmup.lag_threshold`	Final	Dynamic

Use Cases

Disable warmup mid-progress - emergency escape hatch if warmup is taking too long
Adjust lag threshold - tune without recreating the index
Change timeout - increase/decrease without recreating the index

Dynamic Update Behavior

Scenario	Behavior
Disable warmup while in progress (timeout -> -1)	Warmup completes immediately, shard starts serving
Change threshold while warmup in progress	New threshold takes effect on next check cycle
Change timeout while warmup in progress	New timeout takes effect on next check cycle
Enable warmup after shard is already STARTED	Does NOT re-trigger warmup (would be disruptive)

Settings Propagation Path

Update Settings API -> IngestionEngine.registerDynamicIndexSettingsHandlers() -> DefaultStreamPoller.updateWarmupConfig()

Testing

testUpdateWarmupConfigDisableWhileInProgress - disabling warmup via config update
testUpdateWarmupConfigDisableWithRunningPoller - disabling mid-warmup with running poller, verifies updateWarmupStatus disabled branch
testUpdateWarmupConfigThresholdAndTimeoutWhileInProgress - threshold and timeout update during warmup
testUpdateWarmupConfigDoesNotReEnableAfterCompletion - no re-trigger after shard is serving
testWarmupSkippedWhenPollerStartsInPausedState - paused poller skips warmup
testDynamicWarmupSettingsUpdate (IT) - end-to-end dynamic settings update via API

github-actions · 2026-03-20T03:09:42Z

PR Reviewer Guide 🔍

(Review updated until commit `9dcf919`)

Here are some key observations to aid the review process:

🧪 PR contains tests
🔒 No security concerns identified
✅ No TODO sections
🔀 Multiple PR themes Sub-PR theme: Core dynamic warmup settings implementation Relevant files: server/src/main/java/org/opensearch/cluster/metadata/IndexMetadata.java server/src/main/java/org/opensearch/index/engine/IngestionEngine.java server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java server/src/main/java/org/opensearch/indices/pollingingest/StreamPoller.java Sub-PR theme: Tests for dynamic warmup settings Relevant files: server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java plugins/ingestion-kafka/src/internalClusterTest/java/org/opensearch/plugin/kafka/IngestFromKafkaIT.java
⚡ Recommended focus areas for review Warmup Guard Removed The polling loop condition changed from `if (!warmupComplete && warmupConfig.isEnabled())` to just `if (!warmupComplete)`. This means `updateWarmupStatus()` is now called on every poll cycle even when warmup is disabled at startup (before any dynamic update). When warmup is disabled, `updateWarmupStatus()` immediately sets `warmupComplete = true` and counts down the latch — which is the intended behavior — but it also means the latch is counted down inside the poll loop rather than at initialization. If `awaitWarmupComplete` is called before the poll loop runs even once, it will block until the first poll iteration. Previously, with `warmupConfig.isEnabled()` guarding the call, a disabled warmup would never enter `updateWarmupStatus()` at all, so the latch would never be counted down via that path. This is a behavioral change: callers of `awaitWarmupComplete` on a disabled-warmup poller now depend on the poll loop running at least once. if (!warmupComplete) { updateWarmupStatus(); } Race Condition on warmupConfig `warmupConfig` is declared `volatile`, which ensures visibility of the reference but not atomicity of the read-then-use pattern in `updateWarmupStatus()`. If `updateWarmupConfig()` is called concurrently while `updateWarmupStatus()` is mid-execution (e.g., after reading `warmupConfig.isEnabled()` but before reading `warmupConfig.lagThreshold()`), the two reads could see different `WarmupConfig` instances. While `WarmupConfig` appears to be an immutable record, the volatile reference swap between two reads in the same method invocation is still a potential inconsistency. Assigning `warmupConfig` to a local variable at the start of `updateWarmupStatus()` would eliminate this. if (paused \|\| !warmupConfig.isEnabled()) { warmupComplete = true; warmupLatch.countDown(); logger.info( "Warmup skipped for index {} shard {} - {}", indexName, shardId, paused ? "poller is paused" : "warmup is disabled" ); return; } Misleading Test Assertion In `testUpdateWarmupConfigDisableWhileInProgress`, the comment says "isWarmupComplete() checks warmupConfig.isEnabled(), so disabling makes it return true immediately". However, `isWarmupComplete()` returns the `warmupComplete` field, not a live check of `warmupConfig.isEnabled()`. The test passes only because the poller was never started, so `warmupComplete` remains `false` and the assertion `assertTrue(warmupPoller.isWarmupComplete())` would actually fail unless `isWarmupComplete()` has special logic. If `isWarmupComplete()` does check `warmupConfig.isEnabled()` as a shortcut, this is undocumented behavior that could mislead future maintainers. The test should verify the actual contract clearly. // isWarmupComplete() checks warmupConfig.isEnabled(), so disabling makes it return true immediately assertTrue(warmupPoller.isWarmupComplete());

github-actions · 2026-03-20T03:10:17Z

PR Code Suggestions ✨

Latest suggestions up to 9dcf919

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Fix incorrect warmup completion assertion without starting poller The test comment claims `isWarmupComplete()` checks `warmupConfig.isEnabled()`, but looking at the production code, `isWarmupComplete()` likely returns the `warmupComplete` volatile boolean field directly. Since the poller was never started and `updateWarmupStatus()` was never called, `warmupComplete` would still be `false`. This test assertion may be incorrect and could be testing a behavior that doesn't actually exist, leading to a false-positive or test failure. server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java [1054-1055] -// isWarmupComplete() checks warmupConfig.isEnabled(), so disabling makes it return true immediately +// After disabling warmup, the poller loop will call updateWarmupStatus() which will +// detect warmup is disabled and set warmupComplete=true. Since poller is not started, +// we verify the config was updated but warmupComplete is still false until the loop runs. +assertFalse(warmupPoller.isWarmupComplete()); +// Start poller and wait for warmup to complete via the poll loop +warmupPoller.start(); +assertTrue("Warmup should complete after being dynamically disabled", warmupPoller.awaitWarmupComplete(5000)); assertTrue(warmupPoller.isWarmupComplete()); Suggestion importance[1-10]: 7 __ Why: The test comment and assertion appear incorrect - `isWarmupComplete()` likely returns the `warmupComplete` boolean field directly, not checking `warmupConfig.isEnabled()`. Since the poller is never started in this test, `warmupComplete` would remain `false`, making the `assertTrue` assertion potentially wrong and the test unreliable.	Medium
Possible issue	Guard against null warmup config updates The warmup check no longer guards against calling `updateWarmupStatus()` when warmup is disabled, relying instead on the check inside `updateWarmupStatus()`. However, since `warmupConfig` is now volatile and can be updated at any time, there's a potential issue: if warmup was already complete (`warmupComplete == true`), the outer check correctly skips the call. But if warmup was disabled via `updateWarmupConfig` while `warmupComplete` is still `false`, the latch countdown happens inside `updateWarmupStatus()` on the next iteration, which is correct. The real concern is that `warmupConfig` could be set to `null` if `updateWarmupConfig` is called with a null value, causing a NullPointerException. Add a null check in `updateWarmupConfig`. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [439-448] -if (!warmupComplete) { - updateWarmupStatus(); +@Override +public void updateWarmupConfig(IngestionSource.WarmupConfig newConfig) { + if (newConfig == null) { + throw new IllegalArgumentException("WarmupConfig must not be null"); + } + this.warmupConfig = newConfig; + logger.info( + "Warmup config updated for index {} shard {}: timeout={}, lagThreshold={}", + indexName, + shardId, + newConfig.timeout(), + newConfig.lagThreshold() + ); } Suggestion importance[1-10]: 4 __ Why: Adding a null check in `updateWarmupConfig` is a reasonable defensive measure, but the method is only called from `IngestionEngine.updateWarmupConfig` which always constructs a new `WarmupConfig` object, making a null value unlikely in practice. The improvement is minor.	Low
General	Handle individual warmup setting updates independently The `addSettingsUpdateConsumer` with two settings registers a bi-consumer that fires only when both settings are updated together. If only one setting is updated (e.g., only `lag_threshold`), this handler may not be triggered, leaving the warmup config partially stale. Consider registering separate consumers for each setting that read the current value of the other setting from `engineConfig.getIndexSettings()`. server/src/main/java/org/opensearch/index/engine/IngestionEngine.java [493-497] private void updateWarmupConfig(TimeValue timeout, Long lagThreshold) { if (streamPoller != null) { streamPoller.updateWarmupConfig(new IngestionSource.WarmupConfig(timeout, lagThreshold)); } } +private void updateWarmupTimeout(TimeValue timeout) { + Long lagThreshold = engineConfig.getIndexSettings().getValue(IndexMetadata.INGESTION_SOURCE_WARMUP_LAG_THRESHOLD_SETTING); + updateWarmupConfig(timeout, lagThreshold); +} + +private void updateWarmupLagThreshold(Long lagThreshold) { + TimeValue timeout = engineConfig.getIndexSettings().getValue(IndexMetadata.INGESTION_SOURCE_WARMUP_TIMEOUT_SETTING); + updateWarmupConfig(timeout, lagThreshold); +} + Suggestion importance[1-10]: 7 __ Why: The `addSettingsUpdateConsumer` with two settings registers a bi-consumer that only fires when both settings are updated atomically. If a user updates only `lag_threshold` or only `timeout`, the handler won't be triggered, leaving the warmup config stale. The suggested fix of registering separate consumers for each setting is a valid approach to handle partial updates correctly.	Medium

Previous suggestions

Suggestions up to commit 678aa45

Category	Suggestion	Impact
Possible issue	Immediately complete warmup when dynamically disabled When `updateWarmupConfig` is called with a disabled config (timeout=-1) after warmup is already complete, the `warmupComplete` flag is `true` and the latch is already counted down, so there is no issue. However, if warmup is already complete and a new enabled config is set, the polling loop will never re-enter `updateWarmupStatus()` because `warmupComplete` is `true`. But if warmup is in progress and the config is updated to disabled, the latch countdown only happens on the next poll loop iteration. To ensure immediate effect when disabling warmup, the latch should be counted down directly in `updateWarmupConfig` when the new config is disabled and warmup is not yet complete. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [438-448] @Override public void updateWarmupConfig(IngestionSource.WarmupConfig newConfig) { this.warmupConfig = newConfig; logger.info( "Warmup config updated for index {} shard {}: timeout={}, lagThreshold={}", indexName, shardId, newConfig.timeout(), newConfig.lagThreshold() ); + if (!newConfig.isEnabled() && !warmupComplete) { + warmupComplete = true; + warmupLatch.countDown(); + logger.info("Warmup immediately completed for index {} shard {} - warmup was dynamically disabled", indexName, shardId); + } } Suggestion importance[1-10]: 6 __ Why: This is a valid improvement — when warmup is dynamically disabled via `updateWarmupConfig`, immediately counting down the latch ensures threads blocked on `awaitWarmupComplete` are unblocked without waiting for the next poll loop iteration, which is more responsive and correct behavior.	Low
General	Fix incorrect test assumption about warmup completion The comment states that `isWarmupComplete()` checks `warmupConfig.isEnabled()`, but looking at the production code, `isWarmupComplete()` returns `warmupComplete` field directly (a boolean flag), not `warmupConfig.isEnabled()`. This test may be asserting incorrect behavior — without starting the poller, `updateWarmupStatus()` is never called, so `warmupComplete` remains `false` even after disabling the config. The test assertion may pass only if `isWarmupComplete()` has special logic for disabled configs, which should be verified. server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java [1026-1055] public void testUpdateWarmupConfigDisableWhileInProgress() throws Exception { ... - // isWarmupComplete() checks warmupConfig.isEnabled(), so disabling makes it return true immediately + // Dynamically disable warmup (timeout=-1) - should immediately complete warmup + IngestionSource.WarmupConfig disabledConfig = new IngestionSource.WarmupConfig(TimeValue.timeValueMillis(-1), 100L); + warmupPoller.updateWarmupConfig(disabledConfig); + + // After disabling, warmup should be immediately marked complete assertTrue(warmupPoller.isWarmupComplete()); + warmupPoller.close(); +} + Suggestion importance[1-10]: 5 __ Why: The suggestion raises a valid concern: if `isWarmupComplete()` returns the `warmupComplete` field directly and the poller hasn't started, disabling the config alone may not set `warmupComplete=true`. This could indicate a test that passes for the wrong reason, though the actual behavior depends on the `isWarmupComplete()` implementation not shown in the diff.	Low
General	Restore warmup enabled guard in polling loop The warmup check no longer guards against calling `updateWarmupStatus()` when warmup is disabled, relying instead on the check inside `updateWarmupStatus()`. However, `updateWarmupStatus()` when warmup is disabled will set `warmupComplete = true` and call `warmupLatch.countDown()` on every polling loop iteration until `warmupComplete` is set. Since `warmupComplete` is only set inside `updateWarmupStatus()`, the outer guard should be restored to also check `warmupConfig.isEnabled()` to avoid unnecessary method calls after warmup is already skipped. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [240-242] -if (!warmupComplete) { +if (!warmupComplete && warmupConfig.isEnabled()) { + updateWarmupStatus(); +} else if (!warmupComplete && !warmupConfig.isEnabled()) { updateWarmupStatus(); } Suggestion importance[1-10]: 1 __ Why: The `improved_code` is logically equivalent to the existing code (both branches call `updateWarmupStatus()`), making this suggestion incorrect and unhelpful. The outer `!warmupComplete` guard already prevents repeated calls once warmup is done.	Low

Suggestions up to commit 775437a

Category	Suggestion	Impact
Possible issue	Prevent re-enabling warmup after it has completed When `updateWarmupConfig` is called with a disabled config (timeout=-1) after warmup is already complete, the `warmupComplete` flag is already `true` and the latch is already counted down, so this is safe. However, if warmup is already complete and a new enabled config is passed in, the `warmupComplete` flag remains `true` but `warmupConfig.isEnabled()` now returns `true`, which could cause `isWarmupComplete()` to return inconsistent results if it checks `warmupConfig.isEnabled()`. The method should guard against re-enabling warmup once `warmupComplete` is already `true`. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [438-448] @Override public void updateWarmupConfig(IngestionSource.WarmupConfig newConfig) { + if (warmupComplete) { + logger.info( + "Warmup already complete for index {} shard {}, ignoring config update", + indexName, + shardId + ); + return; + } this.warmupConfig = newConfig; logger.info( "Warmup config updated for index {} shard {}: timeout={}, lagThreshold={}", indexName, shardId, newConfig.timeout(), newConfig.lagThreshold() ); } Suggestion importance[1-10]: 6 __ Why: This is a valid concern - if `updateWarmupConfig` is called with an enabled config after `warmupComplete` is already `true`, it could cause inconsistency in `isWarmupComplete()` if that method checks `warmupConfig.isEnabled()`. The test `testUpdateWarmupConfigDoesNotReEnableAfterCompletion` validates this behavior, but the guard is not in the implementation itself.	Low
General	Guard latch countdown to prevent redundant calls Calling `warmupLatch.countDown()` multiple times is safe (it won't go below zero), but `warmupComplete = true` is set redundantly on every call when warmup is disabled. More importantly, if `updateWarmupConfig` is called to disable warmup while the poller is running, `updateWarmupStatus()` will be called on the next iteration and correctly complete warmup. However, if warmup was previously enabled and `warmupComplete` is already `true`, the outer `if (!warmupComplete)` guard prevents this code from running, so there's no double-countDown issue. The logic is correct but could be made clearer by checking `warmupComplete` before calling `countDown()`. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [460-463] // Skip warmup if poller is paused or warmup is disabled if (paused \|\| !warmupConfig.isEnabled()) { warmupComplete = true; - warmupLatch.countDown(); + if (warmupLatch.getCount() > 0) { + warmupLatch.countDown(); + } Suggestion importance[1-10]: 2 __ Why: `CountDownLatch.countDown()` is already safe to call multiple times (it won't go below zero), and the outer `if (!warmupComplete)` guard in the polling loop prevents this code from being reached after `warmupComplete` is set to `true`. The suggestion adds unnecessary complexity for a non-issue.	Low
General	Avoid redundant warmup status checks after completion The warmup check no longer guards against calling `updateWarmupStatus()` when warmup is disabled, relying instead on the check inside `updateWarmupStatus()`. However, `updateWarmupStatus()` when warmup is disabled will set `warmupComplete = true` and call `warmupLatch.countDown()` on every iteration until `warmupComplete` is set. Since `warmupComplete` is only set inside `updateWarmupStatus()`, the outer guard should be restored to also check `warmupConfig.isEnabled()` to avoid unnecessary method calls once warmup is disabled but before the flag is set. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [240-242] -if (!warmupComplete) { +if (!warmupComplete && warmupConfig.isEnabled()) { + updateWarmupStatus(); +} else if (!warmupComplete) { updateWarmupStatus(); } Suggestion importance[1-10]: 1 __ Why: The `improved_code` is logically equivalent to the existing code - both branches call `updateWarmupStatus()` when `!warmupComplete` is true, making the suggestion a no-op. The outer `!warmupComplete` guard already prevents redundant calls after completion.	Low

Suggestions up to commit 422d33c

Category	Suggestion	Impact
Possible issue	Fix incorrect assertion about warmup completion without running poller The comment claims `isWarmupComplete()` checks `warmupConfig.isEnabled()`, but looking at the production code, `isWarmupComplete()` returns the `warmupComplete` volatile field directly. Disabling warmup via `updateWarmupConfig` only sets the new config — the `warmupComplete` flag is only set to `true` inside `updateWarmupStatus()` which runs in the poll loop. Since the poller is never started in this test, `warmupComplete` will remain `false`, making this assertion incorrect. server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java [1054-1055] -// isWarmupComplete() checks warmupConfig.isEnabled(), so disabling makes it return true immediately -assertTrue(warmupPoller.isWarmupComplete()); +// updateWarmupConfig only updates the config; warmupComplete is set by the poll loop. +// Without starting the poller, warmupComplete remains false. +assertFalse(warmupPoller.isWarmupComplete()); Suggestion importance[1-10]: 7 __ Why: This is a valid and important catch. The test asserts `assertTrue(warmupPoller.isWarmupComplete())` after calling `updateWarmupConfig` without starting the poller, but `isWarmupComplete()` returns the `warmupComplete` volatile field which is only set in the poll loop via `updateWarmupStatus()`. The assertion would fail at runtime, making this a real test correctness bug.	Medium
General	Guard warmup skip logic against redundant state changes The `warmupLatch.countDown()` is called unconditionally even if `warmupComplete` is already `true` and the latch has already been counted down. Since `CountDownLatch` ignores extra `countDown()` calls when already at zero this is safe, but `warmupComplete = true` is also set redundantly. More critically, if warmup was previously enabled and in-progress, then the config is dynamically updated to disable it, the latch will be counted down here — but only on the next poll loop iteration. This is the intended behavior, but the `warmupComplete` flag should be checked first to avoid redundant state changes. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [461-470] if (paused \|\| !warmupConfig.isEnabled()) { - warmupComplete = true; - warmupLatch.countDown(); + if (!warmupComplete) { + warmupComplete = true; + warmupLatch.countDown(); + logger.info( + "Warmup skipped for index {} shard {} - {}", + indexName, + shardId, + paused ? "poller is paused" : "warmup is disabled" + ); + } + return; +} Suggestion importance[1-10]: 4 __ Why: The suggestion is valid — adding a `!warmupComplete` guard avoids redundant assignments and log messages on every poll loop iteration when warmup is already done. However, since `CountDownLatch` safely handles extra `countDown()` calls and `warmupComplete` is a simple boolean write, this is a minor optimization rather than a critical fix.	Low
General	Confirm poll loop warmup check is consistent with new logic The original guard `warmupConfig.isEnabled()` was removed from the poll loop check, and the enabled check was moved inside `updateWarmupStatus()`. However, `updateWarmupStatus()` now sets `warmupComplete = true` and counts down the latch when warmup is disabled. This means if warmup starts enabled and later gets dynamically disabled, the latch is correctly released. But the removal of the outer `isEnabled()` guard means `updateWarmupStatus()` is called on every poll iteration until `warmupComplete` is set — which is fine functionally but the original condition was `!warmupComplete && warmupConfig.isEnabled()`. The new condition `!warmupComplete` is correct since `updateWarmupStatus` handles the disabled case internally. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [240-242] -if (!warmupComplete && warmupConfig.isEnabled()) { +if (!warmupComplete) { updateWarmupStatus(); } Suggestion importance[1-10]: 2 __ Why: The suggestion's `improved_code` is identical to the `existing_code` already in the PR diff, so it adds no change. It merely confirms the existing implementation is correct, which warrants a very low score.	Low

Suggestions up to commit 714a8b5

Category	Suggestion	Impact
Possible issue	Fix test asserting warmup completion without starting poller The test comment claims `isWarmupComplete()` checks `warmupConfig.isEnabled()`, but looking at the code, `isWarmupComplete()` returns the `warmupComplete` field directly. The poller was never started, so `updateWarmupStatus()` was never called to set `warmupComplete = true`. This test may be asserting incorrect behavior — the `warmupComplete` flag would still be `false` since the poller thread never ran. The test should either start the poller and wait, or verify the actual behavior of `isWarmupComplete()`. server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java [1026-1055] public void testUpdateWarmupConfigDisableWhileInProgress() throws Exception { - ... - // isWarmupComplete() checks warmupConfig.isEnabled(), so disabling makes it return true immediately + // Create a poller with warmup enabled + IngestionSource.WarmupConfig enabledConfig = new IngestionSource.WarmupConfig(TimeValue.timeValueMinutes(5), 100L); + DefaultStreamPoller warmupPoller = new DefaultStreamPoller( + new FakeIngestionSource.FakeIngestionShardPointer(0), + fakeConsumerFactory, + "", + 0, + partitionedBlockingQueueContainer, + StreamPoller.ResetState.NONE, + "", + errorStrategy, + StreamPoller.State.NONE, + 1000, + 1000, + 10000, + indexSettings, + new DefaultIngestionMessageMapper(), + enabledConfig + ); + + // Warmup should not be complete yet + assertFalse(warmupPoller.isWarmupComplete()); + + // Dynamically disable warmup (timeout=-1) + IngestionSource.WarmupConfig disabledConfig = new IngestionSource.WarmupConfig(TimeValue.timeValueMillis(-1), 100L); + warmupPoller.updateWarmupConfig(disabledConfig); + + // Start the poller so updateWarmupStatus() is called and warmupComplete is set + warmupPoller.start(); + + // Wait for warmup to complete after being disabled + assertTrue("Warmup should complete after being dynamically disabled", warmupPoller.awaitWarmupComplete(5000)); assertTrue(warmupPoller.isWarmupComplete()); + warmupPoller.close(); +} + Suggestion importance[1-10]: 7 __ Why: The suggestion raises a valid concern: `isWarmupComplete()` returns the `warmupComplete` field directly, and since the poller is never started in `testUpdateWarmupConfigDisableWhileInProgress`, `updateWarmupStatus()` is never called to set `warmupComplete = true`. The test assertion at line 1055 may be passing for the wrong reason or could be incorrect, making this a potentially significant test correctness issue.	Medium
General	Restore warmup enabled guard in polling loop The warmup check no longer guards against calling `updateWarmupStatus()` when warmup is disabled, relying instead on the check inside `updateWarmupStatus()`. However, `updateWarmupStatus()` when warmup is disabled will set `warmupComplete = true` and call `warmupLatch.countDown()` on every poll loop iteration until `warmupComplete` is set. Since `warmupComplete` is only set inside `updateWarmupStatus()`, and the outer check was removed, the first call will set it, but the condition should still short-circuit. The original guard `!warmupComplete` is still present, so this is fine — but if warmup is dynamically re-enabled after completion (warmupComplete=true), the `updateWarmupStatus()` will never be called again. Consider restoring the `warmupConfig.isEnabled()` check in the outer condition to avoid unnecessary calls when warmup is disabled. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [240-242] -if (!warmupComplete) { +if (!warmupComplete && warmupConfig.isEnabled()) { + updateWarmupStatus(); +} else if (!warmupComplete) { updateWarmupStatus(); } Suggestion importance[1-10]: 2 __ Why: The suggestion's `improved_code` is logically equivalent to just `if (!warmupComplete) { updateWarmupStatus(); }` since both branches call `updateWarmupStatus()`. The actual concern about unnecessary calls is already handled by the `!warmupComplete` guard, and the re-enable scenario is intentionally unsupported per the test. The suggestion doesn't improve correctness.	Low

Suggestions up to commit 1256a3a

Category	Suggestion	Impact
General	Replace fragile sleep with reliable latch wait Using `Thread.sleep(2000)` to wait for the poll loop to iterate is fragile and can cause flaky tests in slow CI environments. Replace the fixed sleep with a polling loop or use `awaitWarmupComplete` with a reasonable timeout to reliably wait for the latch to be counted down. server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java [1090-1099] // Dynamically disable warmup - updateWarmupStatus will handle completion on next poll loop warmupPoller.updateWarmupConfig(new IngestionSource.WarmupConfig(TimeValue.timeValueMillis(-1), 100L)); -// isWarmupComplete() returns true immediately via config check, but we need to verify -// the poll loop also calls updateWarmupStatus() which counts down the latch. -// Wait briefly for the poll loop to iterate and hit the disabled branch. -Thread.sleep(2000); +// Wait for the poll loop to iterate and count down the latch via updateWarmupStatus +assertTrue("Warmup latch should be counted down", warmupPoller.awaitWarmupComplete(5000)); -// Verify the latch was counted down by updateWarmupStatus (not just short-circuited by isWarmupComplete) -assertTrue("Warmup latch should be counted down", warmupPoller.awaitWarmupComplete(0)); - Suggestion importance[1-10]: 7 __ Why: Using `Thread.sleep(2000)` is a well-known source of flaky tests in CI environments. Replacing it with `awaitWarmupComplete(5000)` is a cleaner and more reliable approach that directly tests the intended behavior without an arbitrary fixed delay.	Medium
	Guard against silently re-enabling completed warmup When `updateWarmupConfig` is called with a disabled config (timeout=-1) after warmup is already complete, the `warmupComplete` flag is `true` and the latch is already counted down, so this is safe. However, if a new enabled config is set after warmup was previously completed (re-enabling warmup), the `warmupComplete` flag remains `true` and warmup will never re-run. The method should log a warning or explicitly document/enforce that re-enabling warmup after completion has no effect, to prevent silent misbehavior. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [438-448] @Override public void updateWarmupConfig(IngestionSource.WarmupConfig newConfig) { + if (warmupComplete && newConfig.isEnabled()) { + logger.warn( + "Warmup config update ignored for index {} shard {} - warmup already completed. Re-enabling warmup has no effect.", + indexName, + shardId + ); + return; + } this.warmupConfig = newConfig; logger.info( "Warmup config updated for index {} shard {}: timeout={}, lagThreshold={}", indexName, shardId, newConfig.timeout(), newConfig.lagThreshold() ); } Suggestion importance[1-10]: 4 __ Why: The suggestion correctly identifies that re-enabling warmup after completion has no effect due to the `warmupComplete` flag, and adding an explicit warning log improves observability. The test `testUpdateWarmupConfigDoesNotReEnableAfterCompletion` already validates this behavior, but the warning log would make the silent no-op more visible to operators.	Low
	Restore early exit when warmup is already complete The warmup check no longer guards against calling `updateWarmupStatus()` when warmup is disabled, since the `warmupConfig.isEnabled()` check was moved inside `updateWarmupStatus()`. However, if `warmupConfig` is updated to a disabled state while `warmupComplete` is already `true`, the latch has already been counted down and this is fine. But if warmup was never started (disabled from the beginning), `updateWarmupStatus()` will be called on every poll loop iteration unnecessarily until it sets `warmupComplete = true`. Consider restoring the `warmupConfig.isEnabled()` guard in the outer check to avoid redundant calls when warmup is disabled from the start. server/src/main/java/org/opensearch/indices/pollingingest/DefaultStreamPoller.java [240-242] -if (!warmupComplete) { +if (!warmupComplete && warmupConfig.isEnabled()) { + updateWarmupStatus(); +} else if (!warmupComplete && !warmupConfig.isEnabled()) { updateWarmupStatus(); } Suggestion importance[1-10]: 1 __ Why: The `improved_code` is logically equivalent to the existing code - both branches call `updateWarmupStatus()` when `!warmupComplete`, making the suggestion a no-op. The concern about redundant calls is valid but minor since `updateWarmupStatus()` sets `warmupComplete = true` on the first call when disabled, preventing further calls.	Low

github-actions · 2026-03-20T04:21:00Z

❌ Gradle check result for 97fffc2: FAILURE

Please examine the workflow log, locate, and copy-paste the failure(s) below, then iterate to green. Is the failure a flaky test unrelated to your change?

github-actions · 2026-03-20T04:54:00Z

Persistent review updated to latest commit 397bc0b

github-actions · 2026-03-20T05:41:34Z

❌ Gradle check result for 397bc0b: FAILURE

Please examine the workflow log, locate, and copy-paste the failure(s) below, then iterate to green. Is the failure a flaky test unrelated to your change?

github-actions · 2026-03-20T06:11:11Z

Persistent review updated to latest commit 2fdf911

Adds testUpdateWarmupConfigDisableWithRunningPoller which starts a poller with high lag (warmup never completes on its own), then dynamically disables warmup and verifies it completes via the updateWarmupStatus() disabled branch. This covers the previously uncovered 'warmup is disabled' path in updateWarmupStatus() and the updateWarmupConfig() method. Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

- Combine testUpdateWarmupConfigThresholdWhileInProgress and testUpdateWarmupConfigTimeoutWhileInProgress into single test that updates both values (varunbharadwaj) - Revert assertBusy in testWarmupPhase back to direct assertEquals since warmup guarantees docs are indexed before completion (varunbharadwaj) Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

- Fix testWarmupPhase: check polled count metric (proves warmup processed all messages) then use waitForSearchableDocs (handles async refresh) - Add testDynamicWarmupSettingsUpdate IT: verifies dynamic settings update via update settings API works end-to-end (covers IngestionEngine path) - Add testWarmupSkippedWhenPollerStartsInPausedState: covers the paused branch in updateWarmupStatus (poller starts paused, warmup is skipped) Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Wait for poll loop to execute updateWarmupStatus() with disabled config, covering the '!warmupConfig.isEnabled()' branch (lines 461-470). Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

totalProcessedCount is incremented in a separate processor thread and may lag behind the poller state. Only assert totalPolledCount which is synchronous with the poller state transition. Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

github-actions · 2026-03-24T01:11:45Z

Persistent review updated to latest commit 422d33c

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

github-actions · 2026-03-24T02:05:57Z

Persistent review updated to latest commit 775437a

server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java

github-actions · 2026-03-24T02:11:49Z

Persistent review updated to latest commit 678aa45

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

github-actions · 2026-03-24T02:18:24Z

Persistent review updated to latest commit 9dcf919

varunbharadwaj

Looks good to me. Thanks!

github-actions · 2026-03-24T03:28:33Z

❕ Gradle check result for 9dcf919: UNSTABLE

Please review all flaky tests that succeeded after retry and create an issue if one does not already exist to track the flaky failure.

…ject#20936) Make warmup settings dynamic for pull-based ingestion --------- Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com> Signed-off-by: Gagan Dhakrey <gagandhakrey@Gagans-MacBook-Pro.local>

kaustubhbutte17 requested review from a team, Bukhtawar, CEHENKLE, Rishikesh1159, anasalkouz, andrross, ashking94, cwperks, dbwiddis, gbbafna, jed326, kotwanikunal, mch2, msfroh, owaiskazi19, reta, sachinpkale, saratvemulapalli, shwetathareja and sohami as code owners March 20, 2026 03:07

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from ee35484 to 97fffc2 Compare March 20, 2026 03:07

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch 2 times, most recently from d4f3eec to 397bc0b Compare March 20, 2026 04:52

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from 397bc0b to 2fdf911 Compare March 20, 2026 06:09

kaustubhbutte17 added 12 commits March 24, 2026 06:39

Retrigger CI

31f58f2

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Retrigger CI

acf7455

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Improve coverage for dynamic warmup disable branch in updateWarmupStatus

c6ec2d8

Wait for poll loop to execute updateWarmupStatus() with disabled config, covering the '!warmupConfig.isEnabled()' branch (lines 461-470). Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Retrigger CI

bfd36cc

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Retrigger CI

88c9241

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Retrigger CI

13f2740

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Retrigger CI

67b1f6d

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

Retrigger CI

86ced39

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from 714a8b5 to 422d33c Compare March 24, 2026 01:10

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from 422d33c to 542fb18 Compare March 24, 2026 02:04

Replace Thread.sleep with awaitWarmupComplete in test

775437a

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from 542fb18 to 775437a Compare March 24, 2026 02:04

varunbharadwaj reviewed Mar 24, 2026

View reviewed changes

server/src/test/java/org/opensearch/indices/pollingingest/DefaultStreamPollerTests.java Show resolved Hide resolved

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from 678aa45 to fcf52b6 Compare March 24, 2026 02:16

Retrigger CI

9dcf919

Signed-off-by: Kaustubh Butte <kaustubhbutte17@gmail.com>

kaustubhbutte17 force-pushed the feature/warmup-dynamic-settings branch from fcf52b6 to 9dcf919 Compare March 24, 2026 02:16

varunbharadwaj added the skip-changelog label Mar 24, 2026

varunbharadwaj approved these changes Mar 24, 2026

View reviewed changes

varunbharadwaj merged commit 6082a2a into opensearch-project:main Mar 24, 2026
41 of 51 checks passed

kaustubhbutte17 mentioned this pull request Apr 1, 2026

Add warmup settings documentation for pull-based ingestion opensearch-project/documentation-website#12187

Merged

Conversation

kaustubhbutte17 commented Mar 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Settings Changed

Use Cases

Dynamic Update Behavior

Settings Propagation Path

Testing

Uh oh!

github-actions bot commented Mar 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Reviewer Guide 🔍

(Review updated until commit 9dcf919)

Uh oh!

github-actions bot commented Mar 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Code Suggestions ✨

Previous suggestions

Uh oh!

github-actions bot commented Mar 20, 2026

Uh oh!

github-actions bot commented Mar 20, 2026

Uh oh!

github-actions bot commented Mar 20, 2026

Uh oh!

github-actions bot commented Mar 20, 2026

Uh oh!

github-actions bot commented Mar 24, 2026

Uh oh!

github-actions bot commented Mar 24, 2026

Uh oh!

Uh oh!

github-actions bot commented Mar 24, 2026

Uh oh!

github-actions bot commented Mar 24, 2026

Uh oh!

varunbharadwaj left a comment

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Mar 24, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

kaustubhbutte17 commented Mar 20, 2026 •

edited

Loading

github-actions bot commented Mar 20, 2026 •

edited

Loading

(Review updated until commit `9dcf919`)

github-actions bot commented Mar 20, 2026 •

edited

Loading