Refactoring task cache manager for forecasting

[kaituo]

Description

This PR extracts shared components from ADTaskCacheManager into TaskCacheManager to support the requirements of the forecasting feature. To enable the extraction, this PR also did the following refactoring.

Renamings:

• Method-level in ADTaskCacheManager:

  • addDeletedDetector to addDeletedConfig
  • addDeletedDetectorTask to addDeletedTask
  • hasDeletedDetectorTask to hasDeletedTask
  • pollDeletedDetector to pollDeletedConfig
  • pollDeletedDetectorTask to pollDeletedTask

• Variable-level in AnomalyDetectorSettings:

  • CHECKPOINT_MAINTAIN_QUEUE_MAX_HEAP_PERCENT to AD_CHECKPOINT_MAINTAIN_QUEUE_MAX_HEAP_PERCENT
  • CHECKPOINT_READ_QUEUE_MAX_HEAP_PERCENT to AD_CHECKPOINT_READ_QUEUE_MAX_HEAP_PERCENT
  • CHECKPOINT_SAVING_FREQ to AD_CHECKPOINT_SAVING_FREQ
  • CHECKPOINT_TTL to AD_CHECKPOINT_TTL
  • CHECKPOINT_WRITE_QUEUE_MAX_HEAP_PERCENT to AD_CHECKPOINT_WRITE_QUEUE_MAX_HEAP_PERCENT
  • COLD_ENTITY_QUEUE_MAX_HEAP_PERCENT to AD_COLD_ENTITY_QUEUE_MAX_HEAP_PERCENT
  • DEDICATED_CACHE_SIZE to AD_DEDICATED_CACHE_SIZE
  • ENTITY_COLD_START_QUEUE_CONCURRENCY to AD_ENTITY_COLD_START_QUEUE_CONCURRENCY
  • ENTITY_COLD_START_QUEUE_MAX_HEAP_PERCENT to AD_ENTITY_COLD_START_QUEUE_MAX_HEAP_PERCENT
  • EXPECTED_COLD_ENTITY_EXECUTION_TIME_IN_MILLISECS to AD_EXPECTED_COLD_ENTITY_EXECUTION_TIME_IN_MILLISECS
  • FILTER_BY_BACKEND_ROLES to AD_FILTER_BY_BACKEND_ROLES
  • MAX_ENTITIES_PER_QUERY to AD_MAX_ENTITIES_PER_QUERY
  • MAX_MODEL_SIZE_PER_NODE to AD_MAX_MODEL_SIZE_PER_NODE
  • MAX_MULTI_ENTITY_ANOMALY_DETECTORS to AD_MAX_HC_ANOMALY_DETECTORS
  • MAX_RETRY_FOR_END_RUN_EXCEPTION to AD_MAX_RETRY_FOR_END_RUN_EXCEPTION
  • MAX_SINGLE_ENTITY_ANOMALY_DETECTORS to AD_MAX_SINGLE_ENTITY_ANOMALY_DETECTORS
  • MODEL_MAX_SIZE_PERCENTAGE to AD_MODEL_MAX_SIZE_PERCENTAGE
  • PAGE_SIZE to AD_PAGE_SIZE
  • RESULT_WRITE_QUEUE_MAX_HEAP_PERCENT to AD_RESULT_WRITE_QUEUE_MAX_HEAP_PERCENT

• Class-level:

  • ADRealtimeTaskCache renamed to RealtimeTaskCache

• Package-level:

  • Shifted from org.opensearch.ad.breaker to org.opensearch.timeseries.breaker

Migrations:

• Variables transferred from AnomalyDetectorSettings to TimeSeriesSettings:
  • BATCH_BOUNDING_BOX_CACHE_RATIO
  • CHECKPOINT_MAINTAIN_REQUEST_SIZE_IN_BYTES
  • CHECKPOINT_WRITE_QUEUE_SIZE_IN_BYTES
  • DEFAULT_AD_JOB_LOC_DURATION_SECONDS (renamed to DEFAULT_JOB_LOC_DURATION_SECONDS)
  • ENTITY_REQUEST_SIZE_IN_BYTES
  • HOURLY_MAINTENANCE
  • INTERVAL_RATIO_FOR_REQUESTS
  • LOW_SEGMENT_PRUNE_RATIO
  • MAINTENANCE_FREQ_CONSTANT
  • MAX_COLD_START_ROUNDS
  • MAX_QUEUED_TASKS_RATIO
  • MAX_TOTAL_RCF_SERIALIZATION_BUFFERS
  • MAX_CHECKPOINT_BYTES
  • MEDIUM_SEGMENT_PRUNE_RATIO
  • NUM_MIN_SAMPLES
  • NUM_SAMPLES_PER_TREE
  • NUM_TREES
  • QUEUE_MAINTENANCE
  • RESULT_WRITE_QUEUE_SIZE_IN_BYTES
  • SERIALIZATION_BUFFER_BYTES
  • THRESHOLD_MIN_PVALUE
  • TIME_DECAY

Deletions:

• Obsolete settings and methods:
  • DESIRED_MODEL_SIZE_PERCENTAGE in AnomalyDetectorSettings
  • getDesiredModelSize in MemoryTracker

Modifications:

• MemoryTracker enums renamed for clear differentiation between real-time and historical memory usage, adding REAL_TIME_FORECASTER for a harmonized single-stream and HC analysis approach.

Deprecations

• deprecated the following AnomalyDetectorSettings variables and explained reasons in the comment
  • AD_MAX_RETRY_FOR_UNRESPONSIVE_NODE
  • ENTITY_REQUEST_SIZE_IN_BYTES
  • ENTITY_FEATURE_REQUEST_SIZE_IN_BYTES

Tests:

• Changes validated with a successful Gradle build.

Issues Resolved

#981

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[sudiptoguha]

Good name change

[sudiptoguha]

Good comment -- add "deprecated" section to PR description so that the directional information will be available.

[kaituo]

added

[codecov]

Codecov Report

Merging #982 (7f5a077) into main (0c5b4b9) will increase coverage by 0.12%.
The diff coverage is 92.02%.

❗ Current head 7f5a077 differs from pull request most recent head 3860c77. Consider uploading reports for the commit 3860c77 to get more accurate results

@@             Coverage Diff              @@
##               main     #982      +/-   ##
============================================
+ Coverage     80.37%   80.49%   +0.12%     
- Complexity     4565     4575      +10     
============================================
  Files           333      334       +1     
  Lines         19114    19104      -10     
  Branches       1987     1987              
============================================
+ Hits          15362    15377      +15     
+ Misses         2803     2775      -28     
- Partials        949      952       +3     

Flag	Coverage Δ
plugin	80.49% <92.02%> (+0.12%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
.../java/org/opensearch/ad/ratelimit/BatchWorker.java	100.00% <ø> (ø)
...nsearch/ad/ratelimit/CheckpointMaintainWorker.java	94.73% <ø> (ø)
.../opensearch/ad/ratelimit/CheckpointReadWorker.java	87.26% <ø> (ø)
...opensearch/ad/ratelimit/CheckpointWriteWorker.java	91.25% <ø> (ø)
.../org/opensearch/ad/ratelimit/ConcurrentWorker.java	76.00% <ø> (ø)
...opensearch/ad/ratelimit/EntityColdStartWorker.java	81.57% <ø> (ø)
...nsearch/ad/ratelimit/RateLimitedRequestWorker.java	71.22% <ø> (ø)
...org/opensearch/ad/ratelimit/ResultWriteWorker.java	68.11% <ø> (ø)
...a/org/opensearch/ad/ratelimit/ScheduledWorker.java	93.93% <ø> (ø)
...g/opensearch/ad/ratelimit/SingleRequestWorker.java	76.47% <ø> (ø)
... and 38 more

... and 3 files with indirect coverage changes

[sudiptoguha]

The name is already descriptive, so the comment was not super useful. But is there some other comment that can describe this class? Pitfalls, best-practices (whatever one knows at the moment)? This class can become crucial down the road.

[kaituo]

yeah, added class comment

/**
 * Responsible for tracking and managing the memory consumption related to OpenSearch time series analysis.
 * It offers functionalities to:
 * - Track the total memory consumption and consumption per specific origin.
 * - Monitor reserved memory bytes.
 * - Decide if memory can be allocated based on the current usage and the heap limit.
 * - Estimate the memory size for a ThresholdedRandomCutForest model based on various parameters.
 *
 */

[sudiptoguha]

This seems to be produced by the comparison tool -- but is exactly the same as before?

[kaituo]

it is not exactly the same, but adapted. Changed the comment to

/**
 * Determines if hosting is allowed based on the estimated size of a given ThresholdedRandomCutForest and 
 * the available memory resources.
 * 
 * <p>This method synchronizes access to ensure that checks and operations related to resource availability 
 * are thread-safe.
 *
 * @param configId      The identifier for the configuration being checked. Used in error messages.
 * @param trcf          The ThresholdedRandomCutForest to estimate the size for.
 * @return              True if the system can allocate the required bytes to host the trcf.
 * @throws LimitExceededException If the required memory for the trcf exceeds the available memory.
 *
 * <p>Usage example:
 * <pre>{@code
 * boolean canHost = isHostingAllowed("config123", myTRCF);
 * }</pre>
 */

[sudiptoguha]

Will this be an issue for the claimed million entities?

[kaituo]

Total entities are different from active entities. We can still claim there are million active entities despite this limitation. But we cannot give an accurate total number of entities.

There is a issue tracking this: #302

[sudiptoguha]

Looks good - some q'/clarifications.

[ohltyler]

LGTM

[opensearch-trigger-bot]

The backport to 2.x failed:

The process '/usr/bin/git' failed with exit code 1

To backport manually, run these commands in your terminal:

# Fetch latest updates from GitHub
git fetch
# Create a new working tree
git worktree add .worktrees/backport-2.x 2.x
# Navigate to the new working tree
cd .worktrees/backport-2.x
# Create a new branch
git switch --create backport/backport-982-to-2.x
# Cherry-pick the merged commit of this pull request and resolve the conflicts
git cherry-pick -x --mainline 1 5ac6390e7de643d8177f0a5833995032f3a279ad
# Push it to GitHub
git push --set-upstream origin backport/backport-982-to-2.x
# Go back to the original working tree
cd ../..
# Delete the working tree
git worktree remove .worktrees/backport-2.x

Then, create a pull request where the base branch is 2.x and the compare/head branch is backport/backport-982-to-2.x.