Make GC of repositories fully async for whitelisted namespaces

This change adds a worker to conduct GC on repositories with garbage every 10s.

Fixes #144
This commit is contained in:
Joseph Schorr 2015-06-19 14:55:44 -04:00 committed by Joseph Schorr
parent acd86008c8
commit 70de107268
6 changed files with 111 additions and 10 deletions

30
workers/gcworker.py Normal file
View file

@ -0,0 +1,30 @@
import logging
from apscheduler.schedulers.blocking import BlockingScheduler
from app import app
from data.database import UseThenDisconnect
from data.model.repository import find_repository_with_garbage, garbage_collect_repo
logger = logging.getLogger(__name__)
sched = BlockingScheduler()
@sched.scheduled_job(trigger='interval', seconds=10)
def garbage_collect_repositories():
""" Performs garbage collection on repositories. """
with UseThenDisconnect(app.config):
repository = find_repository_with_garbage(app.config.get('EXP_ASYNC_GARBAGE_COLLECTION', []))
if repository is None:
logger.debug('No repository with garbage found')
return False
logger.debug('Starting GC of repository #%s (%s)', repository.id, repository.name)
garbage_collect_repo(repository)
logger.debug('Finished GC of repository #%s (%s)', repository.id, repository.name)
return True
if __name__ == "__main__":
logging.basicConfig(level=logging.DEBUG)
sched.start()