Skip to content

Conversation

skyemeedan
Copy link
Contributor

@skyemeedan skyemeedan commented Feb 18, 2025

Description

For QA and Live, double number of alegre web server threads to 8 in QA and Live to handle more incoming connections and more appropriately match with allocated container resource (previously were doubling workers processes, but sounds like threads would be more appropriate if we are blocked by io vs processing)

Reference: proposal on https://meedan.atlassian.net/browse/CV2-6066?focusedCommentId=59305

How has this been tested?

This changes the configuration for QA and Live, so has not been tested locally

@skyemeedan skyemeedan marked this pull request as ready for review February 25, 2025 01:27
Copy link
Contributor

@dmou dmou left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

if we're IO bound, should we be looking at using --worker-class=gevent?

@skyemeedan
Copy link
Contributor Author

if we're IO bound, should we be looking at using --worker-class=gevent?

I believe so, but I think we first have to upgrade flask before we can use gevent, which requires updating, which requires removing the text model dependencies, which requires finishing migrating to presto hosted models, which requires these changes to be able to handle existing loads via the new endpoints

@skyemeedan skyemeedan merged commit 6645936 into develop Feb 25, 2025
4 checks passed
skyemeedan pushed a commit that referenced this pull request Mar 3, 2025
commit 008f4f2
Author: ahmednasserswe <ahmed.nasser.swe@gmail.com>
Date:   Fri Feb 28 06:58:11 2025 +0100

    CV2-5077-Add-test-coverage-for-OpenAI-module-in-Alegre (#491)

    * create test_openai.py and add `test_openai_get_document_body`
    * Update test_openai.py with `test_retrieve_openai_embeddings_handles_api_error` and`test_retrieve_openai_embeddings_calls_openai_api`
    * Update test_openai.py to mock redis cache
    * Update test_openai.py to include tests that call the api
    * mocking cache in `test_openai_get_document_body` and `test_retrieve_openai_embeddings_calls_openai_api`
    * asserting openai key is set during CI
    * adding a unit test to find if openai_api_key is in env
    * added test OPENAI key for CI to encoded .env file https://meedan.atlassian.net/wiki/spaces/ENG/pages/1942945794/How+to+encrypt+and+decrypt+.enc+files+in+Alegre

    ---------

    Co-authored-by: Skye Bender-deMoll <skye@meedan.org>

commit d63cc85
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Thu Feb 27 15:48:30 2025 -0800

    update to show checkweb direct dependency on checkApi

commit 60be6ac
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Wed Feb 26 14:39:37 2025 -0800

    added links to new docs in README, moved png files from toplevel into /doc

commit 3636375
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Wed Feb 26 14:28:44 2025 -0800

    added explanation and API dependencies

commit cb7e5b1
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Wed Feb 26 14:22:51 2025 -0800

    renamed and corrected image links

commit 9412b7a
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Wed Feb 26 12:46:38 2025 -0800

    [CV2-5885] first sketch of infra based on notes from Devin

commit 7b135a2
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Thu Feb 27 16:16:11 2025 -0800

    adding --threads back in

commit 4ea58a1
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Thu Feb 27 15:09:00 2025 -0800

    [CV2-6066] doubling worker processes from 8 to 16, and removing --threads since docs show it is only for gevent

commit 94dd470
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Thu Feb 27 11:12:59 2025 -0800

    doubling threads again from 8 to 16

commit 227fa79
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Thu Feb 27 11:11:42 2025 -0800

    disabling some tests to simplify testing protocol

commit 1afea6a
Author: Skye Bender-deMoll <skye@meedan.org>
Date:   Thu Feb 27 11:03:58 2025 -0800

    [CV2-6066] added a load test for async endpoint

commit cf3e2c2
Author: Martin Peck <51542678+sonoransun@users.noreply.github.com>
Date:   Wed Feb 26 08:25:46 2025 -0700

    Fix ecs deploy dependencies with updated version pins. (#493)

    * Try to resolve dependency errors in ecs deploy by updating botocore, boto3, and awscli version pins.

commit c0930a9
Author: Devin Gaffney <itsme@devingaffney.com>
Date:   Tue Feb 25 20:10:14 2025 -0500

    CV2-6022 update tests to use non-deprecated URLs (#487)

    * CV2-6022 update tests to use non-deprecated (presto backed) URLs
    * update queries to no longer use old endpoint
    * test for min_es_score fix
    * add custom error
    * add sleeps, it ended up just being eventual consistency
    * switch to greaters, deprecated test file having weird interaction effects that we probably don't need to solve for right now
    * Update app/test/test_similarity.py

commit 6645936
Author: Skye Bender-deMoll <122867176+skyemeedan@users.noreply.github.com>
Date:   Tue Feb 25 11:00:33 2025 -0800

    [CV2-6066] load tests 3 (#492)

    * [CV2-6066] change warning message to not include actual key value so sentry errors will group
    * double number of alegre web server worker process-THREADS to 8 in QA and Live to handle more connections
    ---------
    Co-authored-by: Skye Bender-deMoll <skye@meedan.org>

commit c4828f5
Author: Devin Gaffney <itsme@devingaffney.com>
Date:   Mon Feb 24 14:32:45 2025 -0800

    [CV2-6075] documentation for Check-Alegre-Presto media indexing and similarity workflows (#489)

    * CV2-6075 initial stab at documentation for Check-Alegre-Presto media indexing and similarity workflows

commit 3494d9e
Author: Skye Bender-deMoll <122867176+skyemeedan@users.noreply.github.com>
Date:   Fri Feb 14 09:06:34 2025 -0800

    [CV2-6066] load tests for text similarity endpoint (#490)

    * adding faker pip requirement for random text generation for benchmarks (testing only)
    * [CV2-6066] draft load testing script for text similarity endpoints
    * updates to benchmark (disabling parallel inserts) and some code comments
    * re-enabled parallel tests, reducing so all tests only run 100 items
    * added check to not run in live
    ---------
    Co-authored-by: Skye Bender-deMoll <skye@meedan.org>

commit 41c9f8e
Author: Skye Bender-deMoll <122867176+skyemeedan@users.noreply.github.com>
Date:   Fri Feb 7 08:45:44 2025 -0800

    [CV2-6020] test updates and handling for missing min_es_score (#485)

    * [CV2-6020] test updates and handling for missing min_es_score
    * added more explicit error messaging and confirmed error status code, corrected failing test
    ---------
    Co-authored-by: Skye Bender-deMoll <skye@meedan.org>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants