ansaurus

Question

When using Parallel Python, is there any way to tell on which machine the job has run?

Answer 1

+1 A:

A uuid1 could help:

>>> import uuid
>>> uuid.uuid1()
UUID('b46fa8cf-1fc1-11df-b891-001641ec3fab')
>>>

See pydoc uuid and the RFC 4122 for more details, I think the last 48 bits are unique to the host. Not sure you you call/return that in Parallel python though.

In the pp.py I found:

self.__stats[hostid] = _Statistics(ncpus, rworker)

Can you then use get_stats() to get at that:

get_stats(self) Returns job execution statistics as a dictionary.

davey 2010-02-22 14:53:20

Unfortunately, I am not sure that this helps. What I want to do, is given a given job find out where it ran. I could do this by modifying the code that is called, which is what I guess you are suggesting, but that is not really what I wanted to do. I was hoping to get the information directly from the returned job in some fashion.

Paul Wagland 2010-02-22 16:31:45

Just as a general comment, when you do an edit, it is normally "polite" to indicate that you have added more information, particularly if it is different in content, rather than just expanded content. Thanks for the new tip, I will try it out tonight.

Paul Wagland 2010-02-24 15:29:28

Sorry, I'd added it as a comment first then wanted to put in the formatted code so I edited the Q and abandoned the comment.

davey 2010-02-25 11:36:07

Davey, updating the answer is definitely the correct thing to do! No need to apologize for that ;-) I was just saying that when you do update the answer, particularly in response to a comment, it is common to put in <b>Edit:</b> before the bit that you are adding.

Paul Wagland 2010-02-26 20:46:34

ansaurus

tags:

views:

answers:

When using Parallel Python, is there any way to tell on which machine the job has run?

related questions