Status of OSG 0.4.1 release

1
Status of OSG 0.4.1 release

• Michael Thomas, Caltech

2
Motivation

• Summarized in Integration/ItbSpec036 twiki
• Major goals
• clean up OSG 0.4.0 documentation and
  configuration
• pre-WS CE stability improvements
• WS-GRAM
• LCG interoperability with demonstrated successes
• GIP fixes and enhancement for storage elements
• Addition of managed fork queue
• Almost all major goals have been met
• This release is a minor upgrade expect more
  changes for OSG 0.6.0

3
Schedule for OSG 0.4.1

• Based on ITB release 0.3.6
• Delivery to provisioning team originally
  scheduled for April 7
• Delayed by two weeks, mostly due to VDT 1.3.10
  release
• Current plans
• VDT 1.3.10b release real soon now
• delivery to provisioning team April 21
• full deployment May 1
• See Integration/ItbRel036Schedule for details

4
Provisioning Assignments

• Documentation (Burt)
• Services validation twiki/web support (Rob)
• Application validations (Ilya?Michael, with
  active guidance by Frank)

5
Documentation

• See Integration/DocumentationTable036 twiki
• 25 documents from 14 contributors
• Most documents are updates/fixes on top of OSG
  0.4.0
• 2 new documents
• ManagedFork (Terrence)
• GumsAdmins (Leigh)
• Most documents were submitted by March 8
• Some require minor corrections
• Instead of twiki?web conversion, this time
  documents will be frozen by changing write access
  to twiki

6
Services Validation

• Virtual integration workshop held on March 22
• 13 ITB sites identified as participants in
  validation
• FNAL (3), Caltech (2), U Chicago, U Iowa, Purdue,
  U Indiana, BNL, U Oklahoma, U Florida, Texas Tech
• Major services validated on most sites at the
  workshop
• ACDC, GridCat, GIP, MonALISA
• GridEx running routinely on the 13 sites
• Status summarized in the validation table
  Integration/SiteValidationTableITB036

7
Application Validations

• 11 VOs participating
• FERMILAB, LIGO, SDSS, CMS, ATLAS, GRASE, CDF,
  DZERO, GADU, NANOHUB, GROW
• Besides, dedicated tests of
• managed fork and srmclient
• Status summarized in Integration/SiteValidationTab
  leITB036
• Many authentication failures
• Exposed gridftp server, globus-url-client, and
  srmclient bugs
• Tests are underway
• table shows results from most VOs
• Others have committed to start participating soon

8
Known Issues

• ManagedFork validation
• forks jobs into the local condor-managed queue
• but lets the administrator control the number of
  such jobs
• optional component
• need to find a representative who will test on
  ITB sites
• susceptible to recent security problem in condor
• srmclient
• comes with OSG CE and WN Client packages, but not
  OSG client
• bug regression causes some configuration to be
  lost in 0.4.1
• fix in progress, will be included in ITB 0.3.7
  cache

9
And more issues...

• Condor security problem
• Affects managed fork
• Update available in upcoming VDT 1.3.10b
• WS-GRAM postponed at request of Globus
• Not ready for deployment support
• Globus-url-copy file permissions
• Update available in upcoming VDT 1.3.10b

10
VDT 1.3.10b

• Fixes for GridFTP server, globus-url-copy client
• File permissions bug
• Multi-level dir creation
• Condor security fix
• Affects managed fork
• WS client added to VDT client
• ITB-0.3.7
• srmclient update

11
Moving forward

• Deploy VDT 1.3.10b as much as possible once
  available
• May not get all sites to participate, hope to get
  10
• Request VOs to revalidate on 1.3.10b sites
• Esp. STAR, ATLAS, CMS for gridftp problems