Virtual Workshop for ITB Rel 0.5

Participants

Ransom, Burt, Alain, Rob Q, Tim C., Steve, Michael, Chris, Neha, Keith, Kristy, Dan, Karthik, Suchandra, Ilya, John W., Xin, Jeff, Tim S. David, Eric, Anand, John R.

If anyone has been missed please add your name.

Background

The goals of this virtual workshop are:

• Install ITB 0.5.2 on a number of sites in advance of the deployment of the next production release of the the OSG 0.6.
• Find problems and provide feedback to VDT.
• Capture experiences and installation/config tips, updating the DocumentationTable for the purposes of provisioning OSG 0.6.

The deliverable of this workshop will be a testbed that will enable further service validations (especially information and accounting services) and application validation by VOs. This workshop will be similar to the one held in March of last year, VirtualWorkshopMar06.

• Release description is available here: http://osg.ivdgl.org/bin/view/ArchivedDocumentation.ITB.ITB050/WebHome (see further for CE install documentation, etc).
• We need volunteers to help with all phases of validation and documentation, in addition to checking the installation at your site.

Coordinates

• January 24-25, 2007 10am-4pm Central
  • Phone: 510-665-5437
  • Meeting ID: 178178
• The meeting will be coordinated by Rob Quick (Wednesday, Jan 24) and Suchandra Thapa (Thursday, Jan 25). Validation oversight will be managed by David Meyers.

Validation

• VTB validation and testing prior to this release is discussed at ValidationTestbed.
• The ValidationPage contains instructions for validating services. The site validation table SiteValidationTableITB052 includes check-off's for passing tests for:
  • VORS
  • CEMon validation via the resource selection service
  • Gratia accounting
  • And more ...
  • As an ITB site administrator, using the site validation table SiteValidationTableITB052 you should check the column labeled: 0.5.2 for your site when your site has passed site_verify.pl and you have set site_status on. This will signal that your site is ready for application level site validation by the participating application validators.

Participating sites

If you intend to particpate in this round of ITB testing and deployment please list your site name here (and describe its resources - head node, compute nodes, job scheduler, and storage) make sure your site has been registered with the GOC.

UC_T2DEV_ITB

• Single gatekeeper host, Condor scheduler, 2 compute nodes, local disk-based storage
• Focus testing with VORS, Gridex, worker-node-client
• Already registered with GOC

CMS-BURT-ITB

• Single gatekeeper, 4 worker nodes w/4 batch slots each, condor scheduler, local disk-based storage
• Already registered with GOC

OUHEP_ITB

• Single gatekeeper, 32 compute nodes (42 CPUs), condor scheduler, local disk-based storage
• Already registered with GOC

BNL_ITB_Test1

• Single gatekeeper, 10 compute nodes, condor scheduler, dCache storage
• Already registered with GOC

FNAL_GPFARM_TEST

• Single gatekeeper, 11 compute nodes, condor scheduler, dCache and legacy storage
• New Gatekeeper fgitb-gk.fnal.gov (GOC was informed today). byebye fnpcg.fnal.gov

FNAL_FERMIGRID_TEST

• Single gatekeeper, forwards to FNAL_GPFARM_TEST, dCache and legacy storage, plus test gums and voms servers
• Already registered with GOC (fgtest1.fnal.gov)

IUPUI-ITB

• Single gatekeeper
• Already registered with GOC

CIT_ITB_1

• Single gatekeeper with 1 compute node and 8 condor batch slots, disk-only storage
• Already registered with GOC

LIGO-CIT-ITB

• Single gatekeeper with 8 worker nodes. Condor scheduler with disk-only storage.
• Registered with GOC

GROW-ITB

• Single gatekeeper with 1 worker nodes. Torque / Maui cluster with disk-only storage.
• Registered with GOC

ITB_INSTALL_TEST_3

• Single gatekeeper with no worker nodes. Condor scheduler with disk-only storage.
• Registered with GOC

IUB-VTB

• Single gatekeeper with no worker nodes. Pre-existing condor scheduler with disk-only storage.
• Registered with GOC

TTU-TESTWULF

• Single gatekeeper with two dual-cpu-dual-core worker nodes. PBS scheduler with local disk for OSG_DATA/READ/WRITE + SRM/DRM SE storage. (Associated SE not yet completely functional.) Using GUMS/PRIMA.
• Registered with GOC

YOUR SITE

• put your info here

Pre-Workshop

• Site registration
  • Registration Instructions from Installation Guide
• Certificates
• Clients

Getting Started

• The DocumentationTable holds the setup of relevant documentation links. Note they are not up to date for the current release. One of the goals of the integration fest is to make a good start at getting these documents up to date, in preparation for OSG 0.6 provisioning and deployment.

Action items and issues

• Need to update DocumentationTable with a minimum of changes in advance of the workshop so that ITB site admins have a good starting point for contributing. Minimal changes made JeffPorter:
  • documents which appear in the installation guide are ordered as such in the table
  • moved some additional sections of the installation guide into separate included documents
  • moved obsolete sections on MIS and MDS to a separate document at end of table

• PBS/LSF Gratia packages need access to log files which the gatekeeper may not allow. This may be something that we want to explore / document in ITB documentation. From Chris Green: I'm not sure what is meant here: if you mean that the files are on a different machine, then Gratia should be installed on the same machine as the log files -- while the VDT packaging and dependencies may constrain Gratia to be on the gatekeeper node, the Gratia software itself does not. The probe should run as root in crontab so should have access to the log files if they are physically available on a connected filesystem. Are they protected by selinux, or something?
• document vdt-control usage
  • JeffPorter: added Starting Service sub-section to the ce install guide
• Update question about updating gums server automatically when gums is installed (preempt and auto answer this) From Suchandra: Added this to the vdt-questions.sh
• review link from jobmanager to jobmanager-fork by default needed to prevent breakage -- SuchandraThapa
• Document SGE Gratia not installed by OSG stack From Chris Green: SGE Gratia has not yet been packaged for VDT.
  • JeffPorter: added comment to the Installing Services subsection that both the web services and Gratia probe for SGE are still in development
• configure_gip overrides ldif modificatins / change configure_gip to merge instead of override ?
• managed-fork returns error, need to specify --server y
• Updated gums documentation to clarify how to run as non-root user (need tomcat5 and apache to be run as non-root and can't use vdt-control for this)
• remove ldap certificate requirement from documention since gris no longer required
• gris should be in optional services list
  • it was decided that documentation should not be (re)added to the installation guide
• document entry must be made in gums config file in order to authorize ce host as user
• use attributes file for old attributes files SuchandraThapa has updated and is testing an enhancement to configure-osg.sh that uses the same mechanism that VDT uses for upgrades.
• configure-osg.sh overwrites osg-attributes.conf file, need to correct
• http://www.lsc-group.phys.uwm.edu/ivdgl/RA/hostcertreq.html needs update for ITB 0.5.2 ?
• bug: osg-user-vo-map.txt crosslinked symlinks after vdt-control is run / update vo package file and ce.package to do the right thing RobQ has updated the VO package and ce package, this is fixed.
• configure-osg: don't start monalisa SuchandraThapa has updated the configure-osg.sh script to remove the instances where services are started using vdt-control.
• need clarification on configure_gip questions
• update Firewall document with current set of port listeners and ports
• make link to vdt-release-notes more prominent in OSG-release notes
• add start/stop services description via vdt-control and enabling non-OSG-standard services via vdt-register-service
  • JeffPorter: added Starting Service sub-section to the ce install guide
• check prominence of documenting need for http service certificate
• update trouble shooting guide with list of logfiles or where to find the list
  • JeffPorter: added Log Files sub-section to the ce install guide trouble shooting document
• gip advertising ldap at 2135 which is no longer present?
• update install guide with checks that site can to do verify different services such as Gratia, CEMon, ...

-- RobGardner - 12 Jan 2007