Skip to content

Make call SchedulerServer::new once in ballista-scheduler process#1537

Merged
alamb merged 3 commits intoapache:masterfrom
Ted-Jiang:issue_1531
Jan 13, 2022
Merged

Make call SchedulerServer::new once in ballista-scheduler process#1537
alamb merged 3 commits intoapache:masterfrom
Ted-Jiang:issue_1531

Conversation

@Ted-Jiang
Copy link
Member

Which issue does this PR close?

Closes #1531 .

avoid tokio::spawn(async move { state_clone.synchronize_job_status_loop(). multi-times

@Ted-Jiang
Copy link
Member Author

@houqp @alamb Plz have a review.

@houqp houqp requested review from alamb and andygrove January 11, 2022 03:40
"Ballista v{} Scheduler listening on {:?}",
BALLISTA_VERSION, addr
);
//there should be only one SchedulerServer in process
Copy link
Contributor

Choose a reason for hiding this comment

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

I don't understand how this code ensures just one SchedulerServer exists per process.

It seems to me what this code does is create a new SchedulerServer on a port that the operating system picks.

Copy link
Contributor

Choose a reason for hiding this comment

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

Perhaps using something like lazy_static (https://docs.rs/lazy_static/1.4.0/lazy_static/, already a dependency) might be worth looking into

Copy link
Member Author

Choose a reason for hiding this comment

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

Thanks, sorry for my wrong annotation. it should be only call SchedulerServer::new() once in process ,
is this right ?

Copy link
Member Author

Choose a reason for hiding this comment

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

its ok to clone() SchedulerServer in multi-thread without call tokio::spawn(async move { state_clone.synchronize_job_status_loop().await });

Copy link
Member

Choose a reason for hiding this comment

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

SchedulerServer::new creates the scheduler loop that runs in the background, that's the thing that's causing the issue.

Copy link
Member Author

Choose a reason for hiding this comment

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

@alamb Could you review it again, Thanks!

Copy link
Contributor

Choose a reason for hiding this comment

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

Sorry -- I don't really know what this code is doing, so I just had an uninformed question; Sorry if this delayed merging unnecessarily.

@Ted-Jiang Ted-Jiang requested a review from alamb January 12, 2022 13:53
@alamb
Copy link
Contributor

alamb commented Jan 13, 2022

If it fixes your issues @Ted-Jiang then it sounds good to me 👍

@alamb alamb merged commit cf76969 into apache:master Jan 13, 2022
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.

Ballista scheduler dead loop in loadtest

3 participants