airflow-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (AIRFLOW-3001) Accumulative tis slow allocation of new schedule
Date Tue, 11 Sep 2018 02:25:00 GMT

    [ https://issues.apache.org/jira/browse/AIRFLOW-3001?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16610026#comment-16610026
] 

ASF GitHub Bot commented on AIRFLOW-3001:
-----------------------------------------

ubermen opened a new pull request #3874: [AIRFLOW-3001] Add task_instance table index 'ti_dag_date'
URL: https://github.com/apache/incubator-airflow/pull/3874
 
 
   [ Description ]
   There was no index composed of dag_id and execution_date. So, when scheduler find all tis
of dagrun like this "select * from task_instance where dag_id = 'some_id' and execution_date
= '2018-09-01 ...'", this query will be using ti_dag_state index (I was testing it in mysql
workbench. I was expecting 'ti_state_lkp' but, it was not that case). Perhaps there's no problem
when range of execution_date is small (under 1000 dagrun), but I had experienced slow allocation
of tis when the dag had 1000+ accumulative dagrun. So, now I was using airflow with adding
new index ti_dag_date (dag_id, execution_date) on task_instance table. I have attached result
of my test :)
   
   [ Test ]
   models.py > DAG.run
   jobs.py > BaseJob.run
   jobs.py > BackfillJob._execute
   jobs.py > BackfillJob._execute_for_run_dates
   jobs.py > BackfillJob._task_instances_for_dag_run
   models.py > DagRun.get_task_instances
   tis = session.query(TI).filter(
   TI.dag_id == self.dag_id,
   TI.execution_date == self.execution_date,
   )
   ![image](https://user-images.githubusercontent.com/6738941/45285016-fb9ecc00-b51c-11e8-945c-c28d81aece02.png)
   ![image](https://user-images.githubusercontent.com/6738941/45285019-fe012600-b51c-11e8-91fa-a66c2293ca5d.png)
   
   
   ### Jira
   
   - [ ] My PR addresses the following [Airflow Jira](https://issues.apache.org/jira/browse/AIRFLOW/)
issues and references them in the PR title. For example, "\[AIRFLOW-XXX\] My Airflow PR"
     - https://issues.apache.org/jira/browse/AIRFLOW-XXX
     - In case you are fixing a typo in the documentation you can prepend your commit with
\[AIRFLOW-XXX\], code changes always need a Jira issue.
   
   ### Description
   
   - [ ] Here are some details about my PR, including screenshots of any UI changes:
   
   ### Tests
   
   - [ ] My PR adds the following unit tests __OR__ does not need testing for this extremely
good reason:
   
   ### Commits
   
   - [ ] My commits all reference Jira issues in their subject lines, and I have squashed
multiple commits if they address the same issue. In addition, my commits follow the guidelines
from "[How to write a good git commit message](http://chris.beams.io/posts/git-commit/)":
     1. Subject is separated from body by a blank line
     1. Subject is limited to 50 characters (not including Jira issue reference)
     1. Subject does not end with a period
     1. Subject uses the imperative mood ("add", not "adding")
     1. Body wraps at 72 characters
     1. Body explains "what" and "why", not "how"
   
   ### Documentation
   
   - [ ] In case of new functionality, my PR adds documentation that describes how to use
it.
     - When adding new operators/hooks/sensors, the autoclass documentation generation needs
to be added.
   
   ### Code Quality
   
   - [ ] Passes `git diff upstream/master -u -- "*.py" | flake8 --diff`
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


> Accumulative tis slow allocation of new schedule
> ------------------------------------------------
>
>                 Key: AIRFLOW-3001
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-3001
>             Project: Apache Airflow
>          Issue Type: Improvement
>          Components: scheduler
>    Affects Versions: 1.10.0
>            Reporter: Jason Kim
>            Assignee: Jason Kim
>            Priority: Major
>
> I have created very long term schedule in short interval. (2~3 years as 10 min interval)
> So, dag could be bigger and bigger as scheduling goes on.
> Finally, at critical point (I don't know exactly when it is), the allocation of new task_instances
get slow and then almost stop.
> I found that in this point, many slow query logs had occurred. (I was using mysql as
meta repository)
> queries like this
> "SELECT * FROM task_instance WHERE dag_id = 'some_dag_id' AND execution_date = ''2018-09-01
00:00:00"
> I could resolve this issue by adding new index consists of dag_id and execution_date.
> So, I wanted 1.10 branch to be modified to create task_instance table with the index.
> Thanks.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message