Hi all,
Monday at 6 is a bit difficult for me, but I will give a somewhat related talk in the Rucio workshop on Friday morning and I guess some of you will be there too.
Just for some overall background info on what we do in Nordugrid: ARC CE is managing the data transfers for ATLAS jobs rather than the pilot, and it also manages a local cache which is basically an area of a shared file system accessible by all the worker nodes. The ATLAS pilot normally uses tools like gfal-copy to move data but for Nordugrid it uses “mv” because the files are already placed in the pilot’s working directory by ARC CE before the job starts.
I think for xcache the workflow is slightly different because the active party in the transfer is the pilot on the worker node rather than the xcache itself. As far as the pilot is concerned it’s doing a normal xrdcp transfer from some grid storage, it’s just that there is a special configuration to proxy through the xcache and read data from there if it’s already cached. So to answer the original question I’m not sure if a special pilot mover is required for xcache, it’s just a different configuration of xrdcp. But I don’t know too many details of xcache so please correct me if I’m wrong!
Cheers,
David
> On 24 Feb 2018, at 09:04, Yang, Wei <[log in to unmask]> wrote:
>
> Than I guess Monday ~6pm may work. I don’t know about David’s schedule yet. Brian will still be at CERN on Monday evening. My guess is that Brian is more interested in Xcache in general but not that much on ATLAS specific things.
>
> regards,
> --
> Wei Yang | [log in to unmask] | 650-926-3338
>
>
>
>
>
>
>
> On 2/24/18, 12:00 AM, "Andrew Washbrook" <[log in to unmask]> wrote:
>
>> Hi Wei,
>>
>> Thanks for bringing this up. Our flights to Geneva don't get in until
>> 4.20pm on the Monday unfortunately. We'd be happy to have a late meeting
>> on the Monday if that suits (~6pm).
>>
>> Any time Tuesday and Wednesday is fine with us - and we should be able
>> to make Friday the week before if needed.
>>
>> Cheers,
>> Andy.
>>
>> On 24/02/18 03:25, Yang, Wei wrote:
>>> We should try. Depend on the ECDF colleagues, Friday the week before may also be an option.
>>>
>>> regards,
>>> --
>>> Wei Yang | [log in to unmask] | 650-926-3338
>>>
>>>
>>>
>>>
>>>
>>>
>>> On 2/23/18, 6:45 PM, "Brian Bockelman" <[log in to unmask]> wrote:
>>>
>>>> Hi,
>>>>
>>>> Would it be possible for this conversation to happen on Monday so I can participate too?
>>>>
>>>> Thanks,
>>>>
>>>> Brian
>>>>
>>>> Sent from my iPhone
>>>>
>>>>> On Feb 22, 2018, at 5:56 PM, Yang, Wei <[log in to unmask]> wrote:
>>>>>
>>>>> Hi David,
>>>>>
>>>>> Andrej told me in December that ND sites are using ARC cache with ATLAS job. Teng Li @ EDCF is looking at integrating Xcache into ATLAS. I have suggested to them that one possibility is to use a new “site mover” that understand Xcache. Obviously there are other ways. I wonder how you approach this.
>>>>>
>>>>> I believe Teng will likely come to CERN for the site jamboree so I wonder if we could meet and discuss this topic.
>>>>>
>>>>> regards,
>>>>> --
>>>>> Wei Yang | [log in to unmask] | 650-926-3338(O)
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> ########################################################################
>>>>> Use REPLY-ALL to reply to list
>>>>>
>>>>> To unsubscribe from the XCACHE-L list, click the following link:
>>>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XCACHE-L&A=1
>>> ########################################################################
>>> Use REPLY-ALL to reply to list
>>>
>>> To unsubscribe from the XCACHE-L list, click the following link:
>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XCACHE-L&A=1
>>
>>
>> --
>> The University of Edinburgh is a charitable body, registered in
>> Scotland, with registration number SC005336.
>>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XCACHE-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XCACHE-L&A=1
|
