PostgreSQL and SAM3 UI

513/R-068 (CERN)



Show room on map
Two topics to discuss: * Project plan * PostgreSQL * SAM3 UI
Participated at CERN:

Eddie, Lionel, Pablo, Marian,  Julia, Alberto, Nicolo, Robert, Alessandro, Luca (last to arrive)
Remotely: Alessandra
Apologies: Maarten, Stefan, David C.  

Review of Open Tickets from indico.

18 open tickets due to end of April still open. 
  5 of them finished, and the presentation still to do.
  Costin's update on publishing from the experiment might be a topic for a future meeting
  Moving to AI only one machine left , 3 tasks blocked due this
  Meeting on Wednesday about CREAMCE submission
  Several HC tickets need to be reviewed, we will check offline.
  Elastic search we have a one person on this.
  Downtime info still a couple of weeks
  HDFS will be postponed
  Transition SAM by 15th of May
  OPS after EGI still to be decided.
  C5 content should be discussed in section leader meeting. today we have some summary generated from OPS, we need to review.

second topic, Robert -- PostgreSQL for SSB
false friends: need to review all the  sql queries (5k lines!) and check if they are ANSI SQL

--- question at the end of the presentation.
Julia: we know how much effort to translate, but we don't know if postrgresql is performant enough. 
Robert: This will have to be evaluated.

Alberto: If we want to move to another SQL DB (e.g. mariaDB), would  it be a big change?
Robert: It would not be big huge work (order of a week).

Alberto: why postgresql?
Pablo: one reason is because is better integrated with python, another is that now mysql is oracle...

Lionel: I propose to have some tools to check that the SQL code is ANSI SQL compliant 

SAM3 interface
did anyone use it? no.

Pablo pointed out three things:

o) Site downtimes are not propagated correctly. Still to be verified

o) Site grouping:  It comes from the VOFeed. LHCb seems to be missing some parts.
it's on side of LHCb to improve grouping if needed.

grouping is on the experiment side. country could be useful, but it's the experiments.

o) validity of the data: in the past it was 24h, now we are testing with validity 2h. for the first implementation we will keep the 24h, then we change. having 2 profiles is difficult, duplication of metrics

Next meeting: Friday 9th May at 14:00. Topics: MVC framework and caching

There are minutes attached to this event. Show them.