I want to run an example in python, but the celery docs are a little confusing. Where can I find a complete example with clients and server that works?
jdunck has quit
Ergo^ joined the channel
sgviking has quit
sgviking joined the channel
I have celery in ~/Programs/python2.7/lib/python2.7/site-packages/celery, but when I run "python demo.py" I get the error "ImportError: No module named celery". I have python installed in my $HOME. What is wrong with my installation?
asksol
xeon123: does "python -m celery" work?
chance: thanks, fixed
xeon123
asksol: fixed
[o__o] joined the channel
fpk^away is now known as flashingpumpkin
I want that several hosts exchange data between them. How many brokers I need?
host1 <---> host2
davidmiller joined the channel
mher has quit
rrix has quit
rrix joined the channel
ustunozgur has quit
danfairs_ joined the channel
zz_johncosta is now known as johncosta
ustunozgur joined the channel
jamescarr joined the channel
runeh
xeon123: everyone talks to the same broker
jamescarr has quit
jamescarr joined the channel
xeon123
and if I have 3 hosts interconnected in the same broker, I can send a task to a specific host?
asksol
you can if you create a dedicated queue
e.g. celery worker -n myname -Q celery,myname
task.apply_async(args, kwargs, queue='myname')
ustunozgur has quit
danfairs has quit
danfairs_ is now known as danfairs
ojii is now known as ojii|afk
ionelmc joined the channel
ionelmc
asksol: you around?
caleb_ joined the channel
johncosta is now known as zz_johncosta
atom32k is now known as atomekk
jamescarr has quit
jamescarr joined the channel
ustunozgur joined the channel
mike_n joined the channel
ustunozgur has quit
ustunozgur joined the channel
sandywalsh has quit
jamescarr has quit
sandywalsh joined the channel
mher joined the channel
dpn`
does setting a rate limit on a task cause other tasks to back up behind it?
even if there are spare workers?
Coalpaw joined the channel
asksol
dpn`: not for messages on the queue
dpn`
asksol: ok doh.. I've got a lot of those ;)
asksol
inside the worker it depends on prefetch limits, e.g. if the prefetch count is 10 it can only hold 10 tasks at a time, so if all of those are rate limited
it will receive the next message on the queue even if idle
dpn`
yeah I've got a few hundred thousand ahead of these :P
asksol
urm
it will *not* receive the next message on the queue even if idle
dpn`
curse me whoring the concurrency to speed up imports
asksol: right, no worries.. i suspected as much
asksol: one quickie while I've got you... memory use ballooning with big chords..
I've disabled the chord for now.. but was wondering if I'm doing something stupid with memory or if it's something you suspect might be a real issue
(ie i'm using chord for a really long chain and i'm just stupid for doing so)
jamescarr joined the channel
ionelmc
asksol: did you try the datagrams?
asksol
udp is useless for this since udp sockets are always writable
dpn`: not sure, but how big is it?
dpn`
asksol: a few hundred thousand items potentially
asksol
dpn`: that is probably not a good idea at this point
ionelmc
asksol: yeah, but you don't need to check for writeable
you just write
dpn`
asksol: ok no problems.. they are shiny though :)
ionelmc
if the worker is available the it will read it
otherwise other workers can read it
asksol
dpn`: have some ideas to make very large chords work, but no time yet
ionelmc: if we don't know if it's writable or not we have to do scheduling manually
dpn`
asksol: ok no worries.. i should have a dig into it so I at least know what you're talking about when it comes to impl details