#celery

asksol: I just wrote a BSON serializer for the presenetation. Do you want it?

is it possible to append data to a message that one want to retry? so that I can evaluate on that data on subsequent retries

Kvisle: Maybe on the retry exception

I'm new to AMQP, maybe I should sanity-check what I'm trying to do;

Kvisle: It has no relation to AMQP but to celery

I understand that.

I have a task I want to perform, consisting of multiple steps ... some place in the middle, I have something that (currently) needs to be performed by a human being ... so I was thinking of creating a task that ensures the existance of a ticket in our ticketing system - and will retry every few minutes until that ticket is resolved ... after that, the task will continue on

You can pass completely different arguments to the same task on retry

ah

then I can just add ticket information

hey all, has anyone use cassandra as backend ? i cant find a sample table definition for cassandra ? problem is i cant save results to the db...

theo_: someone solved that recently, could be in the issues, in the irc logs or on the mailing-list

no idea why it's not in the documentation

the_drow the plan was to document when I used for the project I wrote it for... which never came to fruition =(

hi. is there a way to .get() the result from AsyncResult without removing it from the broker?

i was hoping to build a system that first gets the result, then saves it safely into a database and only after that tells the broker to delete it

mdf, probably you should specify no_ack=False to get() call and ack the message by yourself

hello. I am entering the world of unit testing with celery and wondered if you could recommend good articles about it? or maybe repositories of some packages on github with good real world examples?

until the message is not acked, it shouldn't be removed from the broker afaik

malinoff: yeah, i thought no_ack=False would do that. however, when i tested that by trying to .get() twice, the second .get() didn't get the result by that task id anymore

mdf, AsyncResult instance caches the result received from the first get call, so it should return something

malinoff: ah, the thing is, i made this quick'n'dirty script that tries to get the result from AsyncResult, and i just ran the script twice

mdf, with no_ack=False ?

malinoff: yep

i didn't try other ways of looking at the task queue in between though...

maybe i'll play around a bit and then come back if i can't solve it. thanks for the pointers, malinoff!

mdf, can you run that twice in a single script? Don't use the same AsyncResult instance, for the second .get call create another instance (from celery.result import AsyncResult; res2 = AsyncResult(res1.id); res2.get())

malinoff: ok, i'll try that

more specifically, if one tasks starts a group of tasks, is there any way to deactivate all called chains/groups ... and only test some of the inner variables of that task (or its return value)?

malinoff: ok, that way i get the result from both of them, even with no_ack=True

(and even with a sleep(2) in between the get calls)

mdf, right, because it also caches on the backend side...

It looks like something is acking the task result in your two-dirty-scripts

could it be something from django settings i'm importing?

I don't know :)

well, thanks anyway! i think i know a bit more about where to look at now :)

cool!