[glue-wg] v1.3 documentation, please!

[Maarten.Litmaath at cern.ch]

Hi Paul,

> I'm in the process overhauling how dCache information is published into GLUE: 
> a complete rewrite of how information is propagated.  So I'm looking for 
> definitive information about how to publish Glue through GIP.
> 
> ... and there's the rub.  I can't find any good references for how to provide 
> this information.  I've started a dCache wiki page[1] that holds various 
> nuggets of information, but none of them seem complete or authoritative.
> 
> [1]	http://trac.dcache.org/trac.cgi/wiki/GLUE
> 
> I understand that the current dCache system, is functional through Owen's 
> sterling effort, with various trial-and-error attempts and information 
> conveyed through emails.  I'm looking to avoid this process as much as 
> possible; having accurate information would be a good start!
> 
> First off, the only GLUE v1.3 documentation I could find is from CNAF CVS [2].  
> The most recent version is "Draft 3---16 Jan 2007".  I think Sergio was going 
> to make this the final version, but could this process be clarified?  BTW, 
> there's a typo in the StorageArea, UsedNearlineSize description; there are 
> other sections that look somewhat incomplete (e.g., grep for "TO BE ADDED")
> 
> [2]	
> https://forge.cnaf.infn.it/plugins/scmsvn/viewcvs.php/*checkout*/v_1_3/spec/pdf/GLUESchema.pdf?rev=32&root=glueschema
> 
> Second, I could also find no authoritative source of information on Glue/LDAP 

Might this be a start:

    http://jra1mw.cvs.cern.ch:8180/cgi-bin/jra1mw.cgi/glue-schema/

> binding for v1.3.  There are some CNAF notes[3], but these mention, in bold, 
> that "[t]his version is for early evaluation and is not meant to be deployed 
> yet".  Can someone say what is incorrect on this page?  More importantly, can 
> someone update it so the page is correct?
> 
> [3]	http://glueschema.forge.cnaf.infn.it/SpecV13/LDAP
> 
> The CNAF notes also mention that certain attributes are "deprecated and their 
> use should be removed from any software".  Does this mean 
> info-provides "must" (or "should", or "may", see RFC 2119) refrain from 
> publishing deprecated attributes?  For example, should no info-provider be 
> publishing GlueChunkKey anymore?

That cannot be right!  The GlueVOInfo object needs GlueChunkKey attributes
to allow it to be linked to both the SE and the SA to which it refers!
Examples here:

    https://twiki.cern.ch/twiki/bin/view/LCG/GSSDGLUEProposal

> For publishing SRM spaces, I found a *proposal* for how this should be 
> done[4].  Confusingly, this contradicts the GLUE/LDAP notes[3] as it 
> stipulates that "mds-vo-name = local" must be used, whereas the notes state 
> that mds-vo-name has been removed from the DIT; in practise "mds-vo-name = 

Indeed...  Well, we did not do that.  I suppose it would have caused too
many problems for just a minor version increment.  Others to comment.

> resource" (as a primary document, for feeding into GIP) is what seems to 
> work.

Yes, it should be "mds-vo-name=resource" now.  I have fixed the page.

> [4]	https://twiki.cern.ch/twiki/bin/view/LCG/GSSDGLUEProposal
> 
> Finally, (prodding Laurence) I could not find *any* statement about what 
> format GIP supports as primary input (although I might have missed this).  
> I'd assume this is roughly GLUE v1.3/LDAP, but apparently there's some magic 
> happens with DNs with RDF of "mds-vo-name=resource".  Is this (still) true?  
> Is it required? What other mappings and translations are possible?

A site BDII queries site resources that have "mds-vo-name=resource" (BDII)
or "mds-vo-name=local" (MDS GRIS), and substitutes "resource" or "local"
with the site name.  A top-level BDII _inserts_ "mds-vo-name=local" right
before "o=grid" for various reasons.  That is another matter to improve.