2010-2-18
| 22:30 | asksol | result.ready() |
| 22:30 | asksol | from celery.result import AsyncResult |
| 22:30 | asksol | result = AsyncResult(task_id) |
| 22:30 | rburhum | ah, so I have to save my task_id on my own table |
| 22:30 | rburhum | there is no, "Registry, give me all task ids" |
| 22:31 | asksol | no, there's not. if you need that, you have to implement it yourself |
| 22:31 | asksol | it's optimized around the common case |
| 22:31 | rburhum | got it |
| 22:31 | rburhum | thanks |
| 22:32 | asksol | please share if you implement it in a generic way :) |
| 22:32 | asksol | or not so generic, whatever. could be nice as an example |
| 22:46 | rburhum | sure, i will |
| 22:47 | abadr | Celery is taking up a few hundred MB. Any idea why? |
| 22:47 | abadr | Or how I can find out? Bringing my server to a crawl |
| 22:50 | abadr | kill'd it. wasn't processing any jobs. |
| 22:50 | abadr | would love to find out why that happened |
| 23:00 | asksol | abadr: running with DEBUG on? |
| 23:00 | abadr | asksol: Django? one server yes, one server no |
| 23:01 | asksol | with a few hundred MB, are you talking RAM? |
| 23:01 | abadr | yes |
| 23:01 | abadr | just today it jumped up... didn't even have any jobs |
| 23:02 | abadr | it's been running the background for a week or so |
| 23:02 | asksol | if settings.DEBUG is on, that would probably happen at some point |
| 23:02 | abadr | why? |
| 23:02 | asksol | because it logs queries |
| 23:02 | abadr | hm |
| 23:02 | abadr | ok |
| 23:03 | asksol | there could be other reasons, but I haven't heard about it before |
| 23:03 | abadr | I see that I'm also on an old version. Will upgrade to 1.0 and see if that helps. |
| 23:03 | asksol | right, that would do it then |
| 23:03 | asksol | the old version checks the periodic task table almost every second |
| 23:04 | asksol | == a connection.queries the size of the sun |
| 23:04 | abadr | i see |
| 23:04 | asksol | the new version doesn't use the database for periodic tasks at all |
| 23:04 | abadr | just listens to amqp? |
| 23:04 | asksol | no, there's a new service called celerybeat |
| 23:04 | abadr | I saw those scripts installed to /usr/local/bin |
| 23:05 | asksol | it's a centralized, service so it doesn't have to share state, so there's no chance for race conditions |
| 23:05 | asksol | and also, it doesn't have to poll the table every second :) |
| 23:05 | abadr | am I supposed to install celeryinit to init.d? |
| 23:06 | asksol | no, it's just for creating tables when not using python |
| 23:06 | asksol | read the changelog: http://ask.github.com/celery/changelog.html |
| 23:07 | abadr | k |
| 23:07 | asksol | run celeryd like before, just if you use periodic tasks, you need to run celerybeat |
| 23:07 | asksol | if you don't need periodic tasks, don't run it |
| 23:08 | asksol | and don't run more than one celerybeat servers |
| 23:09 | abadr | can you foresee celery having any problems handling two instances of the same app running on the same machine? |
| 23:09 | asksol | -s |
| 23:09 | abadr | one is in debug, one isn't |
| 23:10 | asksol | not sure |
| 23:10 | asksol | but it's pretty theoretical at this stage, with an older version and no similar reports |
| 23:11 | abadr | yep |
| 23:11 | asksol | You should watch it a bit more closely for a while |
| 23:15 | abadr | asksol: sorry, but I'm a little confused. What's the correct way to daemonize celery now if I don't use periodic tasks? Changelog seems to say celerybeat, you say otherwise :) |
| 23:16 | abadr | yeah I will keep an eye on it |
| 23:30 | asksol | that doesn't have anything to do with periodic tasks. The --detach argument has been removed because it was buggy |
| 23:30 | asksol | and there was no fix in sight |
| 23:30 | asksol | so use start-stop-daemon on debian/bsd |
| 23:30 | asksol | or supervisord on anything else |
| 23:31 | asksol | there's example init scripts here:http://github.com/ask/celery/tree/master/contrib/debian/init.d/ |
| 23:32 | asksol | and example supervisord configuration here: http://github.com/ask/celery/tree/master/contri... |
| 23:32 | asksol | supervisord is pretty great, as it also make sure it's always running |
| 23:33 | abadr | oh, I see. thanks |
| 23:50 | abadr | asksol: start-stop-daemon isn't working. it thinks it's working, and a new pid gets written to the pidfile, but then nothing is left running and nothing is written to the log |
Page 2 of 2
← Previous
(164 total)