-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Regression (v1.7.8) - Increased Time Required to start a dbt Project #22709
Comments
Experiencing significant slowdowns too (~7000 dbt models) after upgrading dagster from Before ( After ( I've tested it with different dagster versions |
@tomsej @jonathanneo If you could link your |
@tomsej if possible could you also run
Your definitions are being loaded using that command in a subprocess - that will help us fully understand what the bottleneck is. Adding |
@gibsondan Sure, here are json files you requested:
Let me know if you need something else. @sryza thanks a lot! Should I test the branch? |
If you're able to test the branch that would be awesome! |
I apologize, unfortunately, I wasn't by my computer. I have tested that branch and it is way better!! Seems even faster than in version 1.7.3! In 1.7.3 running |
## Summary & Motivation Now that specs are the source of truth on an `AssetsDefinition`'s dependencies, the `asset_deps` property is a derived property that's O(N) to compute, instead of its previous constant time. This is a source of performance regressions: #22709. This PR avoids using this property in favor of directly getting dependencies from the asset specs, which is a constant time operation. ## How I Tested These Changes
## Summary & Motivation Now that specs are the source of truth on an AssetsDefinition's dependencies, the asset_deps property is a derived property that's O(N) to compute, instead of its previous constant time. This is a source of performance regressions: #22709. While #22874 moves away from using it internally, users could still be expecting constant time behavior. More background here: #22874 (review) ## How I Tested These Changes
Awesome. That change will be included in this week's release. |
Dagster version
1.7.8
What's the issue?
I have a quite big dbt project (~2500 models). Prior to the 1.7.8 update, the startup process for local development was relatively swift, taking around 20 seconds. However, post-update, the same process now exceeds 1.5 minutes. Issue persists even without custom code (see below), suggesting the slowdown is linked to version 1.7.8:
Definition:
dbt_snowflake.py
I attempted to execute Dagster with the DEBUG log level using the following command:
dagster dev --python-file pipelines/locations/dbt_snowflake/dbt_snowflake_definition.py --code-server-log-level debug --log-level debug
, but it failed to yield additional information regarding the cause of the delay. Could you advise on alternative methods to diagnose the source of this performance issue?The issue with prolonged startup times is not limited to
dev
; it also occurs in the Linux staging environment, where each step experiences substantial delays.Flamegraphs
1.7.7.:
![1 7 7](https://private-user-images.githubusercontent.com/6624716/342923825-eb37b839-8c10-4bf9-8c93-90e13ed76c95.svg?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.Lo2bL3INGOaho_U3DPuPEOmFO09ZinZrg1bmrDQvp3Q)
1.7.10:
![1 7 10](https://private-user-images.githubusercontent.com/6624716/342924086-cf6d5c0b-ccfc-4275-95ce-88474fd7b16b.svg?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.TLQw5WCegNlJcQVcvi6VLtDmDS3CZuaA1BhKHkWLptQ)
Pyinstruments
1.7.7.:
![CleanShot 2024-06-26 at 11 28 25@2x](https://private-user-images.githubusercontent.com/6624716/343096934-eab5cf78-cdc7-4bb7-9a0f-86230ef46e5d.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.EEP_W9FiIN2DvKHvkG5H_RxnDB7Zr6BMLKwSZRKadLo)
1.7.10:
![CleanShot 2024-06-26 at 11 26 07@2x](https://private-user-images.githubusercontent.com/6624716/343096030-a6757fbd-d9eb-4530-bcbe-fefa615c144c.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.200XEZdqaRTovfr7WtUEcy71SsQouy5ijpoN-lZmGaM)
What did you expect to happen?
Performance between versions is the same.
How to reproduce?
No response
Deployment type
Local
Deployment details
No response
Additional information
No response
Message from the maintainers
Impacted by this issue? Give it a 👍! We factor engagement into prioritization.
The text was updated successfully, but these errors were encountered: