DjangoBot: #celery

do i need to disable all consumers first?

I don't think so actually, but I have no idea what happens

celerymon could become celerymanager

the signature is channel.queue_delete(queue, if_unused=False, if_empty=False)

yeah, but in the case of deleting queues, I think celerymon isn't the right place

or, maybe it is, but I don't even want to think about the work to be done to get it there

but note that I have no idea what happens with celeryd if the queue suddenly disappears

at worst it just crashes though

need to restart it

ok fine, getting there......

i deleted all but one, as it wasn't int he settings.py, as I'd removed it

all done! That will teach me to change the queue's

oh another thing Ive notived, shell_plus from django extensions doesn't play nice with celery probably not an issue though

mat__: open up an issue describing what doesn't play well, sounds convenient to have it working

basically shell_plus throws a nasty error saying queues are missing if you have multiple settings files

or is it "create an issue?"

some with different queues

either is right, ill go write it up in 2 mins

is there a rst2text tool?

or sphinx to text?

I know it's text, but I just want to strip the tags that would look funky in an e-mail

s/tags/markup

do you end sentences with an url with punctuation or not? what's the best practice?

i usually omit the punctuation, but it always feels weird

guess most software is able to extract the proper link anyway

asksol: there are a bunch of punctuation rules that are weird for techical stuff. like periods are supposed to go inside parenthesis if parenthesis is at end of sentence. makes no sense

asksol, when the parent celeryd process gets a TERM, is there code in celeryd that catches it and passes it along? i only see the HUP handler.

multiprocessing should take care of that

got it

term just raises systemexit

so I catch that and terminate the pool

asksol: dont know of a rest => text tools

I haven't seen this behaviour before, so there must be something with supervisord, no?

it works well with start-stop-daemon here at least

asksol: with celery 1.0 how do you use TaskSet with a function decorated with @task? Or to use TaskSet do I need to continue to define a class that extends Task?

btw new queue system is running sweetly, is there anything I should avoid in doing a 0.9.8 to 1.0.0 update, or any backwards compatible issues?

asksol, could be, just diggin' around

mat__: nah, it's basically the same

mat__: http://github.com/ask/celery/commits/master

see from the 0.9.8 bump

mostly documentation, and two bug fixes

underseapilot: even when using KILL, the child processes are killed

asksol, is the exit handler in celery? can you point me at that?

the master process is the process leader, so if it dies, the others should die

afaik, to circumvent that the child processes must've setsid()ed

maybe I should verify this in the stevens book though

asksol: sweet will update now then

underseapilot: http://github.com/ask/celery/blob/master/celery...

ah, the except SystemExit, got it.

maybe, try to replace the above finally: to except (SystemExit, KeyboardInterrupt)

are you using 0.8.x?

no, using latest from github

and you're not using --supervised or --detach?

ok

because multiprocessing does not seem to behave well if you fork()

which is why I removed them

anybody here using supervisord at the moment?

asksol: Celery 1.0.0 deployed on all machines, not a single issue (yet... :) )

:)

underseapilot: ianschenck uses it

queues spiked to 600 items, just waiting for workers to catch up

I got example config files from him and someone else

asksol, i'm not so sure term raises systemexit

pip uninstall celery; pip install celery==0.8.2 --- I just don't have enough time to update all the servers today :(

asksol, ok so based on real use. thanks.

underseapilot: quick fix maybe, use INT instead of TERM

if there's an issue with term we should fix it though

yah, will send a diff if i get something working.

it seems you're right, neither system exit, or atexit handlers are called on TEMR

TERM

I guess multiprocessing does clean up on TERM though

but good to know if we need to run additional clean up tasks in the future :)

ah, it's on hacker news: http://news.ycombinator.com/item?id=1115220

asksol: you have some kind of logging on the celery site I guess? Would be interesting to see how much traffic comes in

gz too version 1.0.0 release :)

s/too/to

thanks :)

asksol: strange qu, not just celery can you think why a management command would show up in manage.py help but when I run it say Unknown command: 'celeryd'

hmm

no

:/

must be an import

doing that shell_plus bug report now

asksol: i will include celery in future releases of my project since there are no packages for ubuntu/debian/arch/gentoo

what do i have to pay attention because of your license?

thanks for asking, but I don't know. just be nice :)

it's BSD, so you can pretty much do anything with it

you can also add it as a dependency

which is the most common way to do it

I see you don't have any setup.py ?

even though you can include it in your distribution, it's very inconvenient to manage

but if you expect your users to install it via the distro package manager, I guess you could maintain packages for us

especially you could ask what the hell is going on with the debian ITP: http://bugs.debian.org/cgi-bin/bugreport.cgi?bu...

I don't know how the debian package eco system works, but apparently they have made packages for celery

ooh, seems its coming soon

asksol: ive got problems too with debian packaging, its just friggin complicated

asksol: ill exclude it if packages are released for the main distros

im currently trying to write an install bash script