On 09/12/2016 11:32 AM, Marcus Ebert wrote:
>
>> But, now and again, a document is clearly of sufficient importance to
>> warrant more effort to keep it right. For example, the Approved VOs
>> document is made up-to-date several times per month - whenever VOMS
>> connection or security details change.
>
> It's good that the document always has valid information in there, but why
> is so much effort put into the overall process - update a web page, send
> announcement around, each site has to implement it through some (own)
> mechanism on all machines in time. Why can't we just create centrally an
> RPM with all the information that could be used to automatically update
> each site? Beside having a good and necessary documentation, wouldn't
> this make site availability and maintenance easier?
That would be a great way to do it. It's been tried, at WLCG. They
have released RPM versions for ALICE, ATLAS, CMS and LHCb in the past.
I've actually demonstrated prototypes myself, for small VOs, at
Birmingham HepSysMan a few years back. The time might be right to use
this idea, esp. as Yaim declines. The current process, cumbersome as it
is, has the advantage that it already suits a lot of software baselines.
Yaim input is universally understood by the Grid Admins, even once they
quit using Yaim! Also, you see, a VO is supposed to alert us when it
changes its VOMS credentials. But this often doesn't happen. Hence I
have to check every week to see what has occurred in the portal. You'd
be surprised how many "secret changes" have come to light.
But the main problem with any change in this area is : the security
mechanisms in the Grid software are the most beaten-up modules in the
entire system, as far as I can see. They've really piled change on
change on change in that area. Hence it's a thing I always would handle
with kid gloves - it's very easy to make it fall apart at which point
your site is down!
So, basically, I'm afraid of changing to a new way! I could always
release RPMs as well as maintain the old standard. But would people take
it up?
Cheers,
Ste
--
Steve Jones [log in to unmask]
Grid System Administrator office: 220
High Energy Physics Division tel (int): 43396
Oliver Lodge Laboratory tel (ext): +44 (0)151 794 3396
University of Liverpool http://www.liv.ac.uk/physics/hep/
|