Hi Chris,
this is up to now the best description of the philosophy (or a bit
more modest: tactic) behind
YAIM. It is really good to see this being understood and supported. As
I keep on saying:
YAIM is NOT a replacement for LCFGng ..... and as a readymade solution
most useful for smaller sites.
For larger sites, with their own tradition in fabric management, we see
the YAIM scripts as part of a
non ambiguous description of the configuration. This should ease the
adaptation to the tools used at
these sites.
We have been discussing to use WIKI for the collection of "localization
recipes". An open for all WIKI
seems not to be the solution, some of the local hacks are not always a
good solution and can be abused as a way to hack sites by publishing
config tools that are creating security holes.
We aim at a setup where we have a pool of experienced sysadmins who
review new solutions and then add them to WIKI.
There is a whole list of features that sites support and that should be
described there, here are just a few:
Different network layouts (see FZK)
Worker nodes running through NAT
Different batch systems..
Different storage systems..
Multiple CEs (7.3 and SL3)
All kinds of ways to support MPI .........
It would be really great if you would consider to volunteer to be one
of the reviewer for WIKI entries.
markus
p.s. however, in this specific case (adding more VOs) we need to be a
bit more flexible....
p.p.s. in the tradition of IT-GD-GIS I'll award a motivational sticker
to you (ask Laurence about them).
At the next occasion you can pick it up.
On Feb 17, 2005, at 10:54 AM, Brew, CAJ (Chris) wrote:
> Hi,
>
> I don't really want to make peoples lives hard for them and this isn't
> really in response to this specific request more an argument against
> feature creep in general. So...
>
> Since YAIM is designed to be extensible by the Site Admins, could I
> make
> a plea to keep the stock YAIM as just installing the simplest most
> basic
> site but doing it really well, rather than having it support all sorts
> of extra options and possibly becoming too complex to modify easily.
>
> We could then have somewhere (on the wiki perhaps) that those of us who
> are modifying the scripts to adapt to local site conditions can share
> our scripts and experience.
>
> This would help keep the learning curve for new sites small but also
> make it easier to adapt it by having lots of exemplars to hand.
>
> Just my 2p.
>
> Chris.
>
>> -----Original Message-----
>> From: LHC Computer Grid - Rollout
>> [mailto:[log in to unmask]] On Behalf Of Oliver KEEBLE
>> Sent: 17 February 2005 07:58
>> To: [log in to unmask]
>> Subject: Re: [LCG-ROLLOUT] yaim and multiple edg-mkgridmap
>> auth entries
>>
>> Sure - request noted. You're also welcome to open a 'bug' in the yaim
>> category of the LCG Operations Savannah for feature requests.
>>
>> Oliver Keeble Information Technology Department
>> [log in to unmask] CERN
>> +41 22 76 72360 CH-1211 Geneva 23
>>
>>
>> Lev Shamardin wrote:
>>> Dear all,
>>>
>>> At the moment it is impossible to configure several auth
>> lists for the
>>> edg-mkgridmap usign yaim scripts, but we, for example, need
>> this for our
>>> custom VOs. Is it possible to place this feature into the yaim
>>> "feature requests"?
>>>
>>> With LCFGng we could easely do this with
>>> EXTRA(mkgridmap.auths) rdig
>>> mkgridmap.uri_rdig
>> ldap://lcg64.sinp.msu.ru/ou=users,o=registrar,dc=lcg,dc=org
>>>
>>> --
>>> Lev.
>>
>
>
************************************************************************
*******
Markus Schulz
CERN IT
|