[Go-essp-tech] Action item: reach consensus on handling CMIP5 user requests

[stephen.pascoe at stfc.ac.uk]

Hi,

This was discussed at our internal CMIP5 meeting this morning.  We strongly recommend some sort of centralised issue tracking system and, if no other partner has a system in place that they would recommend,
 we can offer our footprints system and do initial triage of queries.  I have setup a meeting with our help-desk manager today to discuss the practicalities.

Martin gives an excellent summary of what would be involved.  We will probably need others in the federation to help with triage, particularly as time-zone differences will have an impact.  There needs to be guidance notes on to whom queries are initially assigned.  There needs to be a system of following-up queries that are left open.  There needs to be an agreed response time (e.g. 1 working day).

Technology-wise footprints is very similar to a bug tracker.  So, like bugzilla and Jira, we would run the footprints application and have the central help-desk email address but everyone in the federation would be involved in dealing with queries.  The interface unifies email and the web -- any changes to a query in the web interface can be emailed to the assignees.  Any email response by the person making the query is automatically logged and forwarded to the assignees.

I will find out whether it's possible to give a short demo at the next ESGF telco.

Cheers,
Stephen.

---
Stephen Pascoe  +44 (0)1235 445980
Centre of Environmental Data Archival
STFC Rutherford Appleton Laboratory, Harwell Oxford, Didcot OX11 0QX, UK

-----Original Message-----
From: go-essp-tech-bounces at ucar.edu [mailto:go-essp-tech-bounces at ucar.edu] On Behalf Of martin.juckes at stfc.ac.uk
Sent: 02 February 2011 09:35
To: drach1 at llnl.gov; go-essp-tech at ucar.edu; esgf-federation-testing at lists.llnl.gov
Subject: Re: [Go-essp-tech] Action item: reach consensus on handling CMIP5 user requests

Hello,

This sounds like an important topic -- I can just offer a few thoughts on how the BADC system Charlotte mentioned works:


  *   It is a "footprints" system, which is a commercial issue tracking package used through a web interface.
  *   Queries sent to the badc at rl.ac.uk are automatically lodged in the system, and assigned to the help desk staff.
  *   The help desk is staffed in office hours --  queries will be answered or re-assigned within one working day.
  *   Queries which cannot be dealt with by the help desk staff are re-assigned to appropriate members of the BADC staff.
  *   Anyone can view the queries assigned to them and the history of interactions on that query.
  *   For security reasons, we have external access to the interface restricted to a small range of IP addresses.

The fact that all queries are automatically put in a well structured database would clearly be a big plus for a system as complex as ESGF.

To make this work for ESGF we would need:

  *   help desk staff to deal with assigning queries
  *   a system of handling split responsibilities (e.g. BADC data node and PCMDI gateway not communicating each other)
  *   a system of managing the people who would access and deal with the queries -- i..e making sure that when queries are assigned to institute X by the help desk there is someone there to look at them.

regards,
Martin

________________________________
From: go-essp-tech-bounces at ucar.edu [go-essp-tech-bounces at ucar.edu] on behalf of Drach, Bob [drach1 at llnl.gov]
Sent: 01 February 2011 20:39
To: go-essp-tech at ucar.edu; esgf-federation-testing at lists.llnl.gov
Subject: [Go-essp-tech] Action item: reach consensus on handling CMIP5 user requests

All,

One of the action items from today's telecon is to reach a consensus on dealing with CMIP5 user requests and bug reports.

There are several related issues:

With a federated system, users will probably not always be aware of what data node or gateway they are interacting with. Some users are bound to get confused about who to ask for help or request a problem. So it's up to the CMIP5 administrators to help them, preferably by:

        - Relaying the request to the relevant parties. For example, if a user at NCAR retrieves a GISS dataset hosted by the PCMDI gateway, resident on the NASA/NCCS datanode, the user may start by contacting NCAR. Depending on the situation, the appropriate response could be:

            - Forward to PCMDI to fix a gateway issue
            - Forward to NCCS to fix a data node problem
            - Forward to GISS to address a quality control issue, possibly necessitating unpublish/republish at NCCS.

        - Tracking the issue to ensure resolution.


For CMIP3 we relied on a single mailing list to the administrators, each following all user reports and responding as they saw necessary. This approach didn't always work well but at least it spread the workload around.  I don't see the same approach necessarily working for CMIP5, because:


  *   There will be many more users; it may be burdensome to follow all issues for all gateways.
  *   There will be more  administrators, which could result in a chaotic or uncoordinated response.

Charlotte mentioned that BADC uses a Help Desk approach. Is this something that could be copied or expanded for CMIP5 as a whole?

Bob
-- 
Scanned by iCritical.
_______________________________________________
GO-ESSP-TECH mailing list
GO-ESSP-TECH at ucar.edu
http://mailman.ucar.edu/mailman/listinfo/go-essp-tech
-- 
Scanned by iCritical.