Rob is correct. Actual servers can change (adds, drops, name changes, etc).
The only thing that needs to be "registered" in how to get to the site and
that will always be site's redirector. This can be one or more hosts and if
more than one typically registered under a single alias name.
Andy
-----Original Message-----
From: Rob Gardner
Sent: Wednesday, July 25, 2012 6:37 AM
To: Alessandro Di Girolamo
Cc: Julia Andreeva ; [log in to unmask] ;
[log in to unmask]
Subject: Re: question about xrootd topology
Ale,
On Jul 25, 2012, at 8:03 AM, Alessandro Di Girolamo wrote:
> Ok Julia,
>
>
> Rob, for the topology, what we would need:
> final goal:
> - we will ask WLCG to have the xroot service become a real service in OSG
> and GOCDB. Then we ask sites to declare their xroot servers. In this way
> the xroot service will become like an SRM, or a GRIDFTP.
Just to clarify, when we speak of xrootd servers at the site I believe in
this context we mean site-level redirectors, not xrootd data servers.
> step-by-step:
> - since I know that "politics" on this is slow, and then sites don't
> publish correctly, we will need in the meanwhile a first list of xroot
> servers.
> - you send me a list with some basic structure, e.g. SiteName (please
> check AGIS for this http://atlas-agis-dev.cern.ch/agis/site/table_view/),
> and root:endpoint:port/path
> - we include them in AGIS, and we publish them so that people can see them
> - we create a way to declare them in AGIS for the future.
> - when the step of having the xroot service in GOCDB/OIM is done and sites
> have published the services then AGIS collects them from there and
> automatically associate them to the right site.
>
> Ok?
All sounds good to me.
- Rob
>
>
>
> --- --- --- --- --- --- --- --- --- --- ---
> Alessandro Di Girolamo
> CERN IT-ES-VOS
> +41764870532 (16-0532)
> http://www.cern.ch/Diggi
> --- --- --- --- --- --- --- --- --- --- ---
>
> On Jul 25, 2012, at 2:50 PM, Julia Andreeva wrote:
>
>>
>> Great, thanks a lot, Alessandro.
>> Looking forward.
>> Cheers
>> Julia
>>
>>
>> On Wed, 25 Jul 2012, Alessandro Di Girolamo wrote:
>>
>>> Ciao Julia,
>>> I think I have to take care about that, Rob gave me already enough info.
>>>
>>> we have already fixed the MWT2 site, I've to check what we publish into
>>> the xml
>>>
>>> Thanks
>>>
>>> --- --- --- --- --- --- --- --- --- --- ---
>>> Alessandro Di Girolamo
>>> CERN IT-ES-VOS
>>> +41764870532 (16-0532)
>>> http://www.cern.ch/Diggi
>>> --- --- --- --- --- --- --- --- --- --- ---
>>>
>>> On Jul 25, 2012, at 2:27 PM, Rob Gardner wrote:
>>>
>>>> Julia,
>>>>
>>>> I'll attempt to answer one part of your question.
>>>>
>>>> On Jul 25, 2012, at 4:46 AM, Julia Andreeva wrote:
>>>>
>>>>> Dear all,
>>>>>
>>>>> Integrating the xrootd transfers into the WLCG Transfer Dashboard
>>>>> we need to resolve topology.
>>>>>
>>>>> We are trying to resolve domain of the servers and clients to a
>>>>> particular site.
>>>>> As an information source , we are currently using the CMS and ATLAS VO
>>>>> topology descriptors, the same ones which are used for SAM.
>>>>>
>>>>> However there are several cases we are currently not able to resolve.
>>>>> Please, see the list below.
>>>>>
>>>>> Any help and recommendations are very much welcome.
>>>>>
>>>>> Thanks a lot
>>>>>
>>>>> Cheers
>>>>>
>>>>> Julia
>>>>>
>>>>> For ATLAS for the domains below we found more than one site, so the
>>>>> question is, which site should be mapped to this domain for the time
>>>>> being
>>>>>
>>>>>
>>>>
>>>> MWT2 is a three-site Tier 2 federation (served by a single set of
>>>> endpoints) with physical locations separated by 3-5 msec RTT, and
>>>> therefore important to distinguish for federated access.
>>>>
>>>>
>>>>> - {"site_name": "MWT2, MWT2_UC", "domain": "mwt2.org", "vo":"atlas"}
>>>>> - {"site_name": "MWT2, MWT2_UC", "domain": "uchicago.edu",
>>>>> "vo":"atlas"}
>>>>
>>>>
>>>> I am not sure how you'd like to deal with this sub-structure in your
>>>> monitoring. For the time being you could simply lump them together
>>>> under MWT2.
>>>>
>>>> - Rob
>>>>
>>>>
>>>>
>>>>> - {"site_name": "Firefly, Nebraska, Nebraska-Lincoln", "domain":
>>>>> "unl.edu", "vo":"atlas"}
>>>>>
>>>>> We could not find perfect match for the following domains, but there
>>>>> is the best guess in some cases
>>>>>
>>>>> {"site_name": "", "domain": "lal.stratuslab.eu"}
>>>>> I guess this one was used for the CMS cloud tests, should we map it to
>>>>> GRIF-LAL?
>>>>>
>>>>> {"site_name": "", "domain": "icecube.wisc.edu"}
>>>>> {"site_name": "", "domain": "che.wisc.edu"}
>>>>> {"site_name": "", "domain": "chtc.wisc.edu"}
>>>>> {"site_name": "", "domain": "lmcg.wisc.edu"}
>>>>> Can we map all those to GLOW?
>>>>>
>>>>> {"site_name": "", "domain": "fltech-grid3.fit.edu"}
>>>>> ?
>>>>> {"site_name": "", "domain": "ph.ed.ac.uk"}
>>>>> ?
>>>>> {"site_name": "", "domain": "red.hcc"}
>>>>> {"site_name": "", "domain": "red.hcc.unl.edu"}
>>>>> Can we map both to to Nebraska?
>>>>>
>>>>> {"site_name": "", "domain": "xlate.ufl.edu"}
>>>>> Can we map it to Uflorida-HPC?
>>>>>
>>>>> {"site_name": "", "domain": "ct.infn.it"}
>>>>> ?
>>>>> {"site_name": "", "domain": "naf.desy.de"}
>>>>> Can we map it to DESY-HH?
>>>>>
>>>>> {"site_name": "", "domain": "physics.ucsd.edu"}
>>>>> Can we map it to UCSDT2?
>>>>>
>>>>> ########################################################################
>>>>> Use REPLY-ALL to reply to list
>>>>>
>>>>> To unsubscribe from the XROOTD-L list, click the following link:
>>>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>>>>
>>>
>>>
>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
