USAG 05 June 2008
notes by Diana

Attendance
-----------
AP   - Min Tsai
CERN - Maria Dimou, Diana Bosio
FR   - 
GGUS - Helmut Dres, Torsten Antoni, Guenter Grein
IT   - Federico Nebiolo
NE   - Jeff Templon, Ron Trompert
UK/I - Claire Devereux
SW  - Kai Neuffer
JRA1 - Francesco Giacomini

Apologies: Cal Loomis

3. Discussion on EGEE III Milestone MSA1.6 on GGUS assessment

To assist Kai in the writing of the document MSA1.6 on the GGUS assessment.
Maria proposed that the USAG makes a table of content for it. In this
way we can make explicit the criteria for the assessment. 

USAG can have a say on what criteria should be included in the
document, in a similar way of what was done in EGEEII. This is a huge
document, that may be reduced. It is linked from the agenda of the
USAG meeting of April 10. 

At the proposal to use the previous assessment as a blueprint, Torsten
made the remark that it does not make sense to distribute it. The
assessment in EGEEII was closely coupled with the plan, as in the plan
one could find the criteria for its assessment.

maria asked if we wish to do the same for the assessment in
EGEEIII. The EDMS reference is document 726136.

Jeff asked if we were discussing the assessment document or a document
to guide the assessment. Diana replied that it was the former and Jeff
then commented that in this case it seems weird that the plan contains
the assessment. Those should be independent documents. 

It was clarified that the plan did not contain the assessment, but
very precise points on which progress could be measured. In this
sense, the assessment was made much easier and "contained" in the
plan. 

Torsten: we should make the assessment based on the GGUS plan for
EGEEIII. But there is a problem, the plan is due end of June, but I
have no number whatsoever for the effort of the various the
subtasks. Without those it is very hard to plan.  

Maria: pity that the VOs are not here today. Does anyone would like to
propose a criteria that is important for them to assess User Support?
Those criteria should be made available before next meeting as the
milestone document is due during the Autumn so we only have few
meetings left to discuss them.

Jeff and Ron asked for a measure of efficiency. In particular Jeff
asked if one could have statistics on the various status transitions, 
1. new -> assigned to TPM
2. TPM -> assigned to ROC
3. assigned -> in progress in ROC
4. assigned to ROC -> assigned to site
5. in progress -> closure

{ndr. transition 4 happens quite often outside of GGUS, as such GGUS
cannot have statistics on that)

ideally divided by kind of tickets:
1. bug in software (corresponding to all the software SUs)
2. failed SAM test (corresponding to tickets issued by CODs)
3. experiment problem (corresponding to VO related tickets 
4. VO tickets (under VoSupport SU)

Torsten: it would be nice to have also statistics over time. Those
metrics are already collected, it is not a big effort to provide
statistics on them

Maria: Can Kai use this data and draw conclusions?

Kai: Yes. One could also ask big and small VOs about their input.

Maria: We could also use the site survey results (Kai agreed) and I
will also ask the VOS for their input on criteria to be used, but it
would be easier if one had some concrete questions to be answered. 

Kai will compile a questionnaire for the VOs based on the site survey results

Conclusion

Torsten will provide the statistics on GGUS tickets
Kai will analyse the statistics and also compile a questionnaire based
on the site survey results
maria will address NA4 Vo managers and the LHC Vos before the end of
the summer to have time to discuss the input for the document.

We shall keep this item on the agenda for future meetings.

Kai disconnects.


4.4 DISCUSSION ALARMS ROUTING TO SITES PROPOSAL


There is new input on the corresponding SL item in savannah from
Maria, GGUS, Jeff and John Gordon.

https://savannah.cern.ch/support/?103942

After some discussion it was agreed that:
- the e-mail will be signed by GGUS using its own certificate
- the body of the e-mail will contain the DN of the original submitter
- the alarm e-mail will follow a formatted template

Diana noted that concerning the template we should involve the actual
people that will submit this kind of e-mails. 

Maria noted that they were invited to join the meeting, and Jeff
commented that ATLAS is in hot water at the moment and this is
probably the reason why they did not join. 

Guenter proposed to implement what we have no, i.e. the signed e-mail
with the current template as the template can always be changed
later. 

Jeff asked clarification on on field of the template the "update
field" which basically is the field used to indicate if the ticket is
a team ticket or not. 

Jeff started a discussion on how to guide ticket submission, but
decided to move the discussion in a savannah item to propose
additional fields like service endpoint.

Maria: additional fields typically interfere with local ticketing
systems, so please describe them very accurately. 

to wrap up, Maria told that Guenter is going to meet the following day
with the security team of KIT (fka FZK) to discuss how GGUS can send
signed e-mails. 

This opened the discussion on where to store the alarm e-mail address
of sites. For the moment it will be maintained in GGUS, but will
eventually move to the GOCDB. 

Maria: from where can GGUS take the list?

Jeff: LCG Twiki. Will send the link. But Guenter know as he used the
LCG names and not the EGEE ones.

DB: We have a problem. Which names shall we use?

A discussion followed, as the mapping is not one-to-one between LCG
and EGEE names, some T1s are distributed, like NL-T1 which is two
sites, and also some UK T2 is 4 sites. 

Maria: GGUS has to use EGEE names, we stated this explicitly many
times in many meetings. 

Diana: maybe we can rescue the situation. If mapping is always
one-to-many it is enough to have many sites in the GOCDB with the same
alarm e-mail. GGUS will only need to maintain the list that maps the
LCG naming scheme with the EGEE one, which should be stable and fixed,
in case some operators use the LCG name.

Maria wraps up the discussion saying that GGUS will have to maintain:
- the alarms e-mail list until GOCDB takes over
- a list of mapping between the T1 and the EGEE sites.


4.1 DISCUSSION ON ESCALATION BODY FOR TICKETS

Maria: Some users would like to have an escalation authority to
complain about tickets. Maybe a button "escalate" that reassign ticket
to OCC

Diana: But then the ticket disappear from the SU's radar

maria: Right, maybe then the button will just cause an "involve
others" for the OCC. 

Diana: yes, but OCC is not the right body to escalate software
tickets. 

Francesco: In EGEEII developers gave support, but now this is not
possible anymore. Developers cannot answer GGUS tickets, only
savannah bugs. 

Diana: but the unit after all is a mailing list. Maybe we can find a
way to solve this, without giving the burden of opening savannah bugs
and follow them to see if they were accepted or rejected to the TPMs. 

Discussion will be taken offline. Progress to be reported at the next
meeting. 

Points 4.2, 4.3, 5, 6.1 will be reported to next meeting as time is
really short

6.2 TPM training

NE ROC asked Min to be trained in order to join the TPM effort, but
nor Min nor Maria coordinate the training. 

Torsten: yes, we use to offer regular training in EGEEII but the
demand was really low. We now plan to offer training on request. There
is no point in setting up an organizational effort given that the
demand is so low. 

Maria: fine. Would you send an e-mail to the ROC managers list to see
which ROC is interested in being trained for TPM duties?

Torsten: will do

Min: I was not directly involved in training, but we had very good
training at CERN two years ago, the involvement of experienced TPMs is
crucial. 

Torsten will write to the ROC managers list to see what is the demand.


6.1 USAG@EGEE08

Consensus was that it is a good idea. Maria will see if she can obtain
a slot at the beginning or at the end of the week so that she can come
only for few days.


4.3 DISCUSSION (BRIEF) ON THE REMINDER E_MAIL FOR "AGENTS"

Guenter would like to avoid having many kind of different reports and
reach a consensus on a good definition before implementing it.

Diana suggested Claire to contact Matt (the originator of the request) 
to get further requirements on what he would like, and the urgency of
it as if not that urgent maybe the item can be moved to the July
release. 



ACTION LIST

20080506-1:
(Was: #164.14) Provide information on the CIC VO XML dump 
Gilles. Now passed to the CIC developers (give us a name please as Gilles left.
Carry forward
This action dates since the ESC meeting of 24 May 2007. Please remind us 
who still  needs this action done and why. David Bouvet will ask Gilles.
????

20080506-3:
Ask the GOCDB developers about Site alarms' mailing lists inclusion.
Claire
Pending.

20080506-4:
Comment on EGEE II DSA1.7 (FZK) http://edms.cern.ch/document/726263, page 
41.
Torsten
????
200806-1:
Provide statistics on elapsed time between status transitions in GGUS
Torsten
Pending

200806-2:
Write a questionnaire for the VOs based on the site survey results to
solicit criteria of assessment for user support
Kai
Pending

200806-3:
Write to NA4 and LHC VO managers to ask criteria for user support
assessment. 
Maria
Depends on 200806-2.

200806-4:
Keep point about MS1.6 in agenda
Maria

200806-5:
send the link of the LCG twiki where the alarm e-mails for the sites
can be found
Jeff
Done

200806-6:
Report progress on the definition of escalation authority for software
tickets. 
Diana and Maria

200806-7:
Poll the ROC managers list to see who needs to attend the training, so
GGUS can send material, pointers to documentation and organize a
training session if needed.
Torsten
Pending

200806-8:
Ask Matt if the item can go to the July release and to add further
input to the savannah item https://savannah.cern.ch/support/?104039

200806-9:
agenda points 4.1, 4.2, 4.3, and 5 to next meeting
Maria
Pending