sPHENIX discussion of software

[David Morrison <morrison AT bnl.gov>]

Hi computing cognoscenti,

Several things related to sPHENIX computing have been on my mind lately
and I thought I'd share some developments (mostly virtual at this point)
and invite some feedback.

Martin and Chris and I have met a few times over the last six months
with Tom Schlagel (head of ITD) and Andrew Ferguson (ITD Infrastructure
and Operations Manager) to discuss updating their infrastructure.
They've been receptive and pretty positive about this, though I'll admit
we have little to show for that yet.

In the end, I'd like to see ITD provide a bunch of key services in
support of collaboration and to provide the means to do single sign on
authentication to those services.

They are tracking Indico development and will be beta testers for CERN's
v2.0 release.  That version has support for SAML, so we believe this
could eliminate the need for collaborators to have a separate Indico
account and password.

These days services are deployed on VMs – not on dedicated hardware -
and ITD seems willing to "throw up a server" if we express interest.  At
the same time, they strongly favor software (and versions of software)
supported by RedHat so they can get help if there are problems.

They're willing to throw up a server running the latest version of
mediawiki.  That is SAML savvy.  With that, perhaps we could handle
accounts more sensibly on the wiki.

They're willing to run GNU mailman v3.0.  This is not part of RedHat's
bundle (though v2.x is).  The latest version seems to be SAML savvy and
has a well defined API, so one could script interactions with it.
That'll be important as we go to a "CMS-lite" process of physics and
performance note development.  In that model, mailing lists are cheap
and one sets up a brand new mailing list specific to a note once a draft
has been released to the collaboration.  That's very tedious to do by
hand, but should be straightforward to automate with a good API.

I've been talking to Dmitry Arkhipkin about adopting STAR's note system
for storing, indexing and searching physics and performance notes once
they're final.

We're using GitHub for version control on source code and that seems to
be working well.  We're using it for free, which is possible as we allow
all our code to be in a publicly visible repository.  There are couple
of reasons why we might want a version control system that isn't
necessarily visible to everyone in the world.  If we rely on 3rd party
proprietary code (this is the case in online computing), we might not be
allowed to upload that code to such a repo.  And, at some point in the
future, we may want to treat analysis and its related code as "for
collaboration eyes only" until we choose to make it public.  It's
possible, but very expensive, to use GitHub in a not-necessarily-public way.

The RCF is going to set up a server running *GitLab*.  That would solve
some of the visibility and cost issues.  It seems like a very good
solution, perhaps even better than GitHub, but we'll need to try it out
and see what it's really like.

That's all for now,
Dave

-- 
David Morrison  Brookhaven National Laboratory  phone: 631-344-5840
                Physics Department, Bldg 510 C    fax: 631-344-3253
                          Upton, NY 11973-5000  email: dave AT bnl.gov