a task with time.sleep(x) doesn't seem to return instantly
I was under the impressong that any @task would return without blocking
impression*
jlucci has quit
jlucci joined the channel
shalabhaggarwal joined the channel
fayaz joined the channel
shalabhaggarwal has quit
jlucci has quit
shalabhaggarwal joined the channel
anuvrat joined the channel
jorvis
Is there a standard way to specify a path for or look up stderr and stdout for any given task?
mdan
theres CELERYD_HIJACK_ROOT_LOGGER, but i think for granularity of a given task you'll want to setup the logger in the worker
task_prerun signal might also be a good place, if you wanted a different log by task
jorvis
True. And for the case I was thinking about my question actually doesn't make any sense. I was needing it for tasks which run a command via subprocess. That module gives me stdout and stderr definitions that I can then store. I should have though that through more before asking.
alexis`m joined the channel
ianawilson has quit
fxdgear has quit
cyp has quit
MVXA has quit
alexis has quit
niluje has quit
alexis`m is now known as alexis
alexis is now known as Guest23384
cyp joined the channel
MVXA joined the channel
niluje joined the channel
fxdgear joined the channel
ianawilson joined the channel
rickmak has quit
atomekk joined the channel
domino14 has quit
rickmak joined the channel
yopiti joined the channel
yopiti has quit
mher joined the channel
mdan
anyone here on edge?
as in master 3.1?
Cromulent joined the channel
Cromulent has quit
johnraz has quit
MerlinDMC has quit
elarson has quit
mher has quit
rbranson has quit
rbranson joined the channel
droodle joined the channel
anuvrat has quit
MerlinDMC joined the channel
elarson joined the channel
droodle has quit
droodle joined the channel
droodle has quit
droodle joined the channel
elarson has quit
MerlinDMC has quit
MerlinDMC joined the channel
yopiti joined the channel
elarson joined the channel
anuvrat joined the channel
mher joined the channel
Ergo^ joined the channel
johnraz joined the channel
howdy
asksol, is kombu currently in a transitional state?
when I tried to pull from master earlier today, I ended up with workers that didn't want to read from queues.. its possible it's some of my commits not playing nicely with yours, i suppose...
also, must retry always be called from within the failed task? is there no way to use the asyncresult instance and tell it to call itself again with the same arguments? I suppose I could find some way to serialize a subgoal representation of the task on the initial apply_async call...
rickmak has quit
rickmak joined the channel
rickmak_ joined the channel
rickmak has quit
rickmak_ is now known as rickmak
mher has quit
ionelmc joined the channel
maznaz joined the channel
mher joined the channel
WoLpH joined the channel
backjlack has quit
backjlack joined the channel
joerg joined the channel
joerg is now known as Guest30791
yopiti has quit
k_sze[work] has quit
Katafalkas joined the channel
Katafalkas has quit
Katafalkas joined the channel
anuvrat has quit
mher has quit
sgviking joined the channel
mher joined the channel
rickmak has quit
dlitvak joined the channel
dlitvak_ joined the channel
dlitvak has quit
asksol
mdan: upgrade kombu and py-amqp
shalabhaggarwal has quit
Coalpaw joined the channel
bkuberek joined the channel
atom32k joined the channel
yopiti joined the channel
Flonka joined the channel
atomekk has quit
fayaz has quit
bkuberek has quit
mher has quit
mher joined the channel
bkuberek joined the channel
bkuberek has quit
bkuberek joined the channel
italorossi joined the channel
italorossi
Hello all, what's the best way to register tasks that are in another codebase? (currently i'm using @celery.task def task: pass)
asksol
italorossi: just import the other codebase
italorossi: tasks are shared between apps by default
italorossi
i'm using different projects on different machines..
asksol
the worker needs to be able to import the source code...
or you may be able to find some serializer that can transfer code objects
but that is very, very tricky to get right
italorossi
actually the source is in the worker and I have a "frontend" project sending tasks to my worker
I was trying to avoid duplicate code between the "frontend" and the worker