[Ops] Subject: VO client/service HTTP headers

[Mark Taylor]

Dear Ops,

in discussion with Tom Donaldson after his servicemon talk in the
Ops session yesterday, we talked about the conventions for the HTTP
"User-Agent" header that have come up previously in Ops sessions.
Since this is not very well publicised, I wanted to take the
opportunity to point it up here for people who are on either end
of service monitoring/validation operations.

The basic idea is that clients hitting VO services for operational
reasons (e.g. service monitoring), rather than in order to make
"real" data access requests, can/should identify themselves with a header
something like:

   User-Agent: servicemon (ivoa-monitor)

If you're a monitor/validator author, or a service operator who wants
to identify which clients are 'real', you can take a look at the wiki
page where we've written this up:

   https://wiki.ivoa.net/twiki/bin/view/IVOA/UserAgentUsage

Along similar lines, in Groningen Markus gave a talk suggesting
use of the HTTP "Server" header (or similar) to identify VO server
software.  This could come in quite useful when we're trying
to track down the origin of validation issues in deployed services.
DaCHS is already doing this:

   % curl http://dc.g-vo.org/tap/capabilities -s -I
   HTTP/1.1 200 OK
   Server: DaCHS/2.0 twistedWeb/18.9.0
   Date: Thu, 19 Nov 2020 13:15:47 GMT
   Content-Type: text/xml

I haven't checked if other VO server suites are doing the same thing,
but I'd encourage them to do that if they can.

Any further discussion on these topics is very welcome.

Mark

--
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-9288776  http://www.star.bris.ac.uk/~mbt/