1 / 6

LFC consolidation STEP-0

LFC consolidation STEP-0. Simone Campana (on behalf of Many). Today’s intervention. First step toward LFC consolidation at CERN DB backend moved from LCGR to ADCR Modified schema for partitioning LFC frontends: NEW frontend: prod- lfc - atlas.cern.ch (3 hosts behind)

jayden
Download Presentation

LFC consolidation STEP-0

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. LFC consolidationSTEP-0 Simone Campana (on behalf of Many)

  2. Today’s intervention • First step toward LFC consolidation at CERN • DB backend moved from LCGR to ADCR • Modified schema for partitioning • LFC frontends: • NEW frontend: prod-lfc-atlas.cern.ch (3 hosts behind) • OLD prod-lfc-atlas-local.cern.ch DECOMMISSIONED (R.I.P) • LEGACY prod-lfc-atlas-central.cern.ch DECOMMISSIONED (R.I.P) • The old DB accounts will be kept locked for 1 month and then retired.

  3. Preparation • Many consolidation tests carried on in the last months for CERN+SARA (+IN2P3) • Two conclusive tests • Consolidation (CERN+SARA) + Stress (Tracker) on INTR DB some weeks ago • Consolidation (CERN+SARA) + Stress (Tracker) on ADCR DB last week • Outcome: • Approx 12h consolidation time on both INTR and ADCR (the later a bit faster) • Stress Test query rate (bulks of 500 files) • 20 Hz on INTR (with GSI, 1 LFC frontend) • 10 Hz on ADCR (with GSI, 1 LFC frontend) • 12 Hz on ADCR (with GSI, 3 LFC frontend) • 20 Hz on ADCR (without GSI, 3 LFC frontend) • This week we will measure also the time to take a consistency dump

  4. What happened today (and yesterday) • Prior to the intervention • Panda queues for CERN closed yesterday (except CERN-RELEASE for HLT reprocessing) • DDM endpoints blacklisted for R/W and deletion • This morning • At 9:22 ATLAS gave green light to start • At 9:36 the OLD LFC has been stopped • From 9:41 to 10:43 the DB was migrated • At 11:35 the new LFC was fully functional • At 11:47 ATLAS ToA was updated.

  5. Today… after the migration • New LFC endpoint needs to be updated in many services • Refreshing ToA cache • DDM services, DaTRI • Changing entries manually • SchedConfig, (installation service), AK47/drchecker • Update acknowledgements: • DDM at 12:00 (Cedric+Vincent+Puppet) • Installation system at 12:46 (A.DeSalvo) • Panda(Mon) from 12:15 to 13:11 (Ueda) • To be verified • DaTRI (different time zone) • drchecker (?) • BNL SS (Hurricane)

  6. Conclusions: • The preparation phase of LFC consolidation took many months and involved many people and groups at CERN: • ATLAS DBAs, IT-DB, IT-ES, IT-GT, IT-PES, PH-ADP • It took longer than expected (tentative date was end of june) , but I think it was necessary • The LFC consolidation process involves many people • Site Managers of the source site • DBAs + experts in the migration itself • ATLAS experts for post-migration steps • The post-migration could (should ) be done by one person (AMOD?) • Instead of many like today • Each service responsible should prepare a how-to for next time • Testing procedure after merge should be improved • 1 LFC “consolidated” • 14 to go … next one tentatively in 2 weeks (SARA)

More Related