Skip to content
This repository has been archived by the owner on Jan 31, 2020. It is now read-only.

Slow operations should be non-blocking #4

Open
brummett opened this issue Aug 7, 2014 · 5 comments
Open

Slow operations should be non-blocking #4

brummett opened this issue Aug 7, 2014 · 5 comments
Labels

Comments

@brummett
Copy link
Contributor

brummett commented Aug 7, 2014

Creating a database, new and from a template
dropping a database

@brummett
Copy link
Contributor Author

brummett commented Aug 8, 2014

uploading a template file

@brummett
Copy link
Contributor Author

The only possible slow thing now can be creating a DB from a template or a template from a DB

@nnutter
Copy link
Contributor

nnutter commented Feb 28, 2015

I'm wondering if create the DB from the template has some Postgres lock contention, see genome/genome#524, which is causing HTTP timeouts.

@nnutter nnutter added the bug label Feb 28, 2015
@nnutter
Copy link
Contributor

nnutter commented Mar 1, 2015

I increased the timeouts to 30 seconds in genome-env and now I saw 4 out of ~12 model tests get this error:

Got error response 502: Proxy Error at /usr/bin/test-db-database-create line 15

@brummett
Copy link
Contributor Author

brummett commented Mar 2, 2015

The original purpose of this issue was to start using the Mojo event loop in places where long-running handlers (like the old dump-database-load to copy DBs) would block the system from servicing other HTTP requests.

The docs for Mojo::Pg shows how this might work. Look for Mojo::IOLoop->delay.

Do you know if the timeouts we're getting are because it's taking too long to service a new HTTP connection (because it's still busy doing the last one) or to copy the database? The first problem would be fixed with this event loop thing, but not the second.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
Projects
None yet
Development

No branches or pull requests

2 participants