EGEE-III INFSO-RI Enabling Grids for E-sciencE EGEE and gLite are registered trademarks Operational Tools M2 Update James Casey SA1 Management Meeting Barcelona

Enabling Grids for E-sciencE EGEE-III INFSO-RI Summary of milestone timeline 2 We are here…

Enabling Grids for E-sciencE EGEE-III INFSO-RI Pending M1 Features - April 2009 Configuration repositories –First version of Aggregate Topology Provider (ATP)  What resources should I test ? ROC level Nagios based monitoring available –‘SAM Portal’ level of visualization complete Full Nagios testing of all resources in grid running –Used to validate equivalence to SAM 3 DONE

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - Regional Dashboard ‘Regionalized’ dashboards interfaced with regional Nagios Raising alarms based on Nagios Nagios notifications are generated + sent to message bus Lavoisier can pick them up, and insert in DB New regional dashboard can display them –No ‘Alarm DB’ masking done Current regionalized dashboard at IN2P3 cannot display Nagios alerts 4 NOT DONE IN PROGRESS

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - Configuration repositories Metric Description Database MDDB exists, and schema seems now finalized –Many iterations to get it to work with metric store and availability calculator –Profiles are now supported in the MDDB  We have a ‘ROC’ and a ‘Site’ profile No good UI yet for adding new metrics, profiles –We do DB inserts to get us bootstrapped –This is a general statement for most of the configuration repositories – any UI that exists is just in ‘demo’ form 5 IN PROGRESS

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - SLA Calculation Multiple simultaneous availability calculations for a VO Design of availability calculation system is done –Based on profiles and the status changes calculated in the metric store Implementation not started yet by Gridview team 6 NOT DONE

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - GOCDB Write functions for Programmatic Interface (XML over HTTP) available –XML over HTTP API available Prototype interface to new GOCDB4 available First regions deployed communicating with Central project-level GOCDB ‘addDowntime’ added as only needed write method GOCDB distributed by RPM –No public demo instance of central GOCDB available yet Work starting with HGSM developers on region 3 example 7 NOT DONE DEMO DONE

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 – ROC level Nagios Now publishes to new central metric store Submission framework fully uses ATP Central metric store result visualization (SAM Portal/Gridview) Waiting on ATP programmatic interface MyEGEE now at ‘demo’ quality –Need to get into ROC Nagios bundle for testing 8 DONE NOT DONE DEMO

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - QR Reporting Portal Added reports for operations and user support use cases –Some still missing (operations.{1,2}, size.2) Operations Portal now scheduled for release in Dec 09 9 POSTPONED

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - Accounting Tested ActiveMQ transport with some selected sites Patch submitted for gLite certification APEL now just ready for testing phase –Currently internal testing has been done –Not deployed at ay sites yet Need to integrate security configuration into main broker network 10 NOT DONE

Enabling Grids for E-sciencE EGEE-III INFSO-RI M2 - GGUS MSG interface for ticket submission/update available Interfaces defined, messages components written on both side –But no complete ticket flow demoed yet Need to integrate security configuration into main broker network 11 IN PROGRESS

Enabling Grids for E-sciencE EGEE-III INFSO-RI Other work done Gstat 2.0 –Very good progress, ahead of schedule – Operations Dashboard –Work done on new regional dashboard –Complete re-write, which is also for the central component GOCDB –Testing with regions –Chasing down people to turn off APIs 12

Enabling Grids for E-sciencE EGEE-III INFSO-RI My personal viewpoint Perhaps some focus lost –We understood the Nagios/SAM area well, but had not yet defined the areas so well at the start of the project –And didn’t take time to re-evaluate where we were Regional flavoured developments are creeping in –Before we have central components to replace the in-production central components ones –New developments are always more attractive ;( We’re now at the point for most components that we have ‘demo’ quality components –Works on laptop – sort-of –It takes 6-12 months to turn this into production  And we don’t have that much time ! 13

Enabling Grids for E-sciencE EGEE-III INFSO-RI What to do ? Accept our milestones have slipped Focus on delivering a replacement set of products by end of year –In the same model as current components –Ease back on regionalization where necessary Add more people to OAT management level –Tracking progress, project management Make clearer the separation of the EGI futures part from the delivery of something in EGEE-III Get something delivered that works for the ROCs –The work from the OAT teams is good, we just need to get it released, evaluated and into production. 14

Enabling Grids for E-sciencE EGEE-III INFSO-RI Resources Architecture and components Milestone tracking 15

EGEE-III INFSO-RI Enabling Grids for E-sciencE EGEE and gLite are registered trademarks OAT mandate till the end of EGEE III Maite Barroso, SA1 activity leader CERN 23 rd September 2009

Enabling Grids for E-sciencE EGEE-III INFSO-RI To cha nge : Vie w - > Hea der and Foo ter 17 Past OAT’s focus in the first year has been –Distributed/regional architecture –Implementation plan with well defined milestones –distributed monitoring (which was more advanced, and was the first step for a distributed regional support)

Enabling Grids for E-sciencE EGEE-III INFSO-RI To cha nge : Vie w - > Hea der and Foo ter 18 Priorities We know where we are today, late with the milestones (we knew we were optimistic ) We have 7 months to go There are two priorities in the operation tool area for the next months: –Have a solution based on new architecture + components that is functionally equivalent to the current one (and nothing else added) –Deploy one component of this in production, really replacing the central instance, get experience with it. The best candidate today is SAM/Nagios + associated databases with interfaces to the rest of the operation tools central instances

Enabling Grids for E-sciencE EGEE-III INFSO-RI To cha nge : Vie w - > Hea der and Foo ter 19 Future Refocus OAT and operation tools teams Reinforce OAT mgt with more people Split into present (deployment) and future (requirements from NGIs)

Enabling Grids for E-sciencE EGEE-III INFSO-RI To cha nge : Vie w - > Hea der and Foo ter 20 proposal OAT main responsible and chair: James Casey Mandate: reporting to the activity management, sit on both the teams and have the global view of both parallel activities OAT development and deployment (present) Mandate: –revise (update), track milestones defined at the beginning of EGEE-III, track the developments committed for these milestones and deploy them in production (M1, M2, M3 and M4) –Complete the definition of deployment/release procedures, and implement them, including verification criteria, early testers, and early SLAs with operation product teams –Own the ops tool interfaces (and finish its definition) Membership: all operational tools in EGEE Chair: Nick Thackray

Enabling Grids for E-sciencE EGEE-III INFSO-RI proposal OAT advisory (future) Mandate: –Act as advisory committee for all operation tools, replacing the present per-tool advisory committees –do a round of requirements gathering and define new set of milestones for regionalization of operation tools, with input from stakeholders (all ROCs and NGIs), to be implemented in the EGI era (post-M4) –With the egee III experience, propose a set of milestones for the next years, including timeline, interaction between tools and effort estimation –Reference regional implementation Membership: operational tools with dev effort in EGI, NGIs, ROCs, related infrastructure projects, VOs? Chair: wait for EGI bidding result To cha nge : Vie w - > Hea der and Foo ter 21

Enabling Grids for E-sciencE EGEE-III INFSO-RI Comments ? 22