1 / 12

CERN Castor DB Services

CERN Castor DB Services. Nilo Segura Chinchilla Eric Grancher. Outline. Hardware Software Optimiser Upcoming changes. Hardware. Change of HW platform for the main Stagers Old HP machines run out of warranty at the end of November

emma
Download Presentation

CERN Castor DB Services

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. CERN Castor DB Services Nilo Segura Chinchilla Eric Grancher

  2. Outline • Hardware • Software • Optimiser • Upcoming changes

  3. Hardware • Change of HW platform for the main Stagers • Old HP machines run out of warranty at the end of November • New platforms, Dell based, with more Cores (8) and Memory (16GB) • Next : Nehalem blades with 48GB RAM

  4. Software • Red Hat 4.0 2.6.9-78.0.22.ELsmp • Problem with recent 4.0 patches and 5.x • RDBMS 10.2.0.4 + CPU July 2009 + one-off patches. • CPU October soon (delayed to October 20th) • No release date yet for 10.2.0.5, expected in 2010 • CASTOR 11g test service to be upgraded to 11gR2

  5. Optimiser • Found problems with empty tables where the optimizer chooses Index Skip Scan or Fast Full Scan over Unique Index access • Subrequesttodo procedure and other places • Bug 5714944 - CBO may choose INDEX SKIP SCAN instead of INDEX UNIQUE SCAN • Workaround is to Hint the query • Fixed in 10.2.0.5 and 11.x • Occasional problems caused by wrong stats and bind peeking

  6. Coming tasks • Help when/if needed to optimisation of the new Tape Gateway component • Maybe merge SRM and Stager into a single db system • Pending the installation of the new hardware • Less systems to administer • Some tests in RAC mode in ITDC. Further tests will be required.

  7. Gold image and patching • We produce RPM for the rdbms / CRS and EM agent (and a additional component for CRS configuration), descriptive and automated • Alreay 16 rdbms 10.2 RPM produced • Process documented and presented at postC5 • They contain base release, Patchset, jumbo patches, individual patches including security (CPU) • Miguel Marquina does all of the patching for the CASTOR database systems 7

  8. Gold images and patching • For 10.2.0.4, many patches • As of 10.2.0.5 and 11.2 we will use Patch Set Updates (contains the CPU), will require quite a lot of “merge patches” • List of patches tested and deployed at https://twiki.cern.ch/twiki/bin/view/DESgroup/OracleCERNPatchLevels 8

  9. Incidents • BigID • Cursor confusion in some condition when parsing a statement who had already been parsed and for which the execution plan has been aged out of the shared pool. • LCG has been good at identifying cursor confusion issues  (3 major issues since start of 10.2, Oracle has started using mutex) • CRS loop file generation • Post-mortem written. Dedicated monitoring added 9

  10. Kernel • Kernel 2.6.9-89 (RHEL 4.8) instability (CASTOR not affected, still on 2.6.9-78): Oops (kernel panic, linked with Ethernet jumbo frames) 10

  11. Backup Workflow Arash/Ruben

  12. Recovery Workflow Arash/Ruben

More Related