On 3 March 2011 11:43, <[log in to unmask]> wrote:
> I am also doing a splurge now (different people to start with should show up any submission issues, the will allow SAM/Wahid to do the ongoing tests you might want to test the number of streams in the transfers ( in option ) try values of 1,4,10?
I'll stick with the defaults (1) for the initial round of tests, I think.
Just trying to make a suitably sensible script... should go online
sometime soon (I have it logging the transfer time to a different file
so I can get simple statistics out of it)
Sam
> Brian
>
> -----Original Message-----
> From: GRIDPP2: Deployment and support of SRM and local storage management [mailto:[log in to unmask]] On Behalf Of Sam Skipsey
> Sent: 03 March 2011 11:01
> To: [log in to unmask]
> Subject: Re: 2G atlas files placed
>
> Hi Alessandra,
>
> On 3 March 2011 10:51, Alessandra Forti <[log in to unmask]> wrote:
>> Hi Sam,
>>
>> weren't you and Wahid going to do it?
>>
>
> Fair enough.
> I'll get it working sometime this evening - I'm not around this afternoon.
>
> Sam
>
>> cheers
>> alessandra
>>
>> On 03/03/2011 10:43, Sam Skipsey wrote:
>>>
>>> Hi Alessandra,
>>>
>>> On 3 March 2011 10:37, Alessandra Forti<[log in to unmask]> wrote:
>>>>
>>>> Hi Sam,
>>>>
>>>> I think we should do a copy every hour. If we think it is too much doing
>>>> all
>>>> the T2<-> T1 every hour we can do one T1 every hour to/from each T2. In
>>>> this way we will have at least a measure for every T1 every 3 hours.
>>>>
>>> I'd tend to agree, in general, with this.
>>>
>>>> I'd also take care of deleting additional copies systematically since we
>>>> are
>>>> not using any catalogue.
>>>>
>>> That's definitely a good idea.
>>>
>>>> Is that ok with you?
>>>>
>>> Are you going to do this centrally, or do you want us to handle the
>>> GLA transfers, Chris to handle QMUL etc, or?
>>>
>>> (I did some more transfers for RAL and SARA to get better statistics -
>>> our values are about 50MB/s from RAL, and around 5MB/s from SARA,
>>> averaged over 3 transfers each.)
>>>
>>> Sam
>>>
>>>> cheers
>>>> alessandra
>>>>
>>>>
>>>> On 03/03/2011 10:08, Sam Skipsey wrote:
>>>>>
>>>>> Hi Alessandra,
>>>>>
>>>>> On 2 March 2011 17:56, Alessandra Forti<[log in to unmask]>
>>>>> wrote:
>>>>>>
>>>>>> Hi Sam and Wahid,
>>>>>>
>>>>>> as agreed during the meeting I put 2GB files in
>>>>>>
>>>>>>
>>>>>>
>>>>>> srm://dcsrm.usatlas.bnl.gov:8443/srm/managerv2?SFN=/pnfs/usatlas.bnl.gov/atlasscratchdisk/user/uk/sonar-2G
>>>>>
>>>>> Works for me - copies at (around the sonar test value of) 1.1MB/s to
>>>>> GLA. Interestingly, the instantaneous rates that gridftp reports for
>>>>> these transfers are multiples of 512KB/s, which I've not seen before
>>>>> (the 1.1MB/s is an integral over 512,1024, 1536 and 2048 KB/s
>>>>> samples). This implies something about the rate limits on the
>>>>> individual channels bonded to provide our link (probably at the BNL
>>>>> end) - there are at least 4 links, each rate-limited to 512KB/s.
>>>>>
>>>>>>
>>>>>> srm://srm.grid.sara.nl/pnfs/grid.sara.nl/data/atlas//atlasscratchdisk/users/uk/sonar-2G
>>>>>
>>>>> Works for me - copies at (around 8 times Sonar) 4.4MB/s to GLA (only
>>>>> one sample here, so...)
>>>>>
>>>>>>
>>>>>> srm://srm-atlas.gridpp.rl.ac.uk/castor/ads.rl.ac.uk/prod/atlas/scratch/atlasscratchdisk/users/uk/sonar-2G
>>>>>
>>>>> Works for me - copies at (20x ! Sonar) 58MB/s to GLA (only one sample
>>>>> here, so..)
>>>>>
>>>>>>
>>>>>> srm://bohr3226.tier2.hep.manchester.ac.uk/dpm/tier2.hep.manchester.ac.uk/home/atlas/atlasscratchdisk/users/uk/sonar-2G
>>>>>> srm://se03.esc.qmul.ac.uk/atlas/atlasscratchdisk/users/uk/sonar-2G
>>>>>>
>>>>>>
>>>>>> srm://svr018.gla.scotgrid.ac.uk:8446/srm/managerv2?SFN=/dpm/gla.scotgrid.ac.uk/home/atlas/atlasscratchdisk/users/uk/sonar-2G
>>>>>>
>>>>> not tried the copies the other way yet...
>>>>>
>>>>> Sam
>>>>>
>>>>>> Let me know if they are ok for you.
>>>>>>
>>>>>> cheers
>>>>>> alessandra
>>>>>>
>>>>
>>
>>
>
|