Tuesday, April 19, 2016

GOC Service Update - Tuesday, April 26, 2016 at 13:00 UTC

The GOC will upgrade the following services beginning Tuesday, April 26 at 13:00 UTC. The GOC reserves 8 hours in the unlikely event unexpected problems are encountered.

GratiaWeb
Fix for project accounting plots
inclusion of HTML site for Google Charts
inclusion of WLCG reporting link

OIM
Increase the number of Security and Administrative contacts for a resource - release 3.52

All services
Operating system updates; reboots will be required. The usual HA mechanisms will be used, but some services will experience brief outages.

Friday, April 15, 2016

HTCondor 8.4.5 causes problems with partitionable slots

HTCondor 8.4.5 (released in OSG 3.3.11 on Tuesday, April 12) contains a bug that may affect some sites significantly. If your site uses partitionable slots on execute nodes, OSG recommends that you avoid HTCondor 8.4.5, either by skipping the update or reverting to a previous version (e.g., HTCondor 8.4.4).

(Tech details: Jobs that land on partitionable slots will fail to start about 10% of the time. The failed job correctly returns to the queue in Idle state, and HTCondor will continue trying to match and run the job which should complete eventually. But the recycling process adds inefficiencies into the overall system.)

The HTCondor development and OSG Software teams discovered the bug after the OSG release, and plan to release a patched version soon.

Thursday, April 14, 2016

OSG GRAM Support Ending November 2016

As part of the OSG transition from GRAM to HTCondor-CE technology, OSG Operations and Software teams will stop supporting GRAM CE installations in early November. A site that requests support for GRAM or GRAM-based software will be asked to upgrade to HTCondor-CE instead. However, GRAM packages will remain available for sites that accept full responsibility for operation of GRAM.

Toward that goal, OSG CE convenience meta-packages (e.g., osg-ce-condor) will drop GRAM components starting with the April 12 release of OSG 3.3. The change will affect only new installations that use the osg-ce-* packages — OSG will still ship GRAM software, will document how to install GRAM, and will support GRAM in tools like osg-configure.

Other milestones that will likely occur in early 2017:

* Ending support for GRAM CEs at the OSG pilot factories, which will leave many VOs unable to run at GRAM sites
* Dropping GRAM software from the OSG software stack (starting in OSG 3.4.0)
* Removing GRAM-based code from OSG tools like osg-configure


BACKGROUND

An OSG CE is the entry point for the OSG to your local resources. At the heart of the CE is the job gateway software, which accepts incoming jobs, authorizes them, and delegates them to your batch system for execution. These days, OSG jobs come from the OSG factories and are really pilot jobs, which in turn run actual end-user jobs. Since OSG started, the Globus GRAM gatekeeper has managed grid jobs, but starting in May 2014, the OSG added another job gateway option — HTCondor-CE — that is built on core HTCondor technology. For many reasons, OSG has decided to migrate all CEs to HTCondor-CE and is today in the midst of this change.

Since August 2014, new CE installs included both HTCondor-CE and GRAM, and since December 2014, HTCondor-CE is the default job gateway software for a new site. Today, HTCondor-CE is on 60 CEs and is the sole job gateway software on many of them; the list of sites that have already migrated includes some of the largest OSG sites, as well as medium and small ones.

This is a significant technical change for OSG, but the good news is that it is well underway and has been going very smoothly for most sites. If you are still running GRAM, please consider migrating to HTCondor-CE as soon as possible (you can run both for a while, if you like). Then, once your HTCondor-CE is running well, decommission the GRAM software.

Tuesday, April 12, 2016

Reminder about the OSG User School 2016

REMINDER:

The deadline for applications is this Friday, April 15!!!

Please help spread the word, especially through personal connections to people who might benefit from the School. Announcing the Open Science Grid User School 2016! If you could access hundreds, thousands, or even more computers for your scholarly work, what could you do? How could it transform your work? What discoveries might you make? We are seeking applicants for the Open Science Grid (OSG) User School 2016, which takes place 25 - 29 July at the beautiful University of Wisconsin in Madison.

Participants will learn to use high throughput computing (HTC) to harness vast amounts of computing power for research, applicable to nearly any field of study (e.g., physics, chemistry, engineering, life sciences, earth sciences, agricultural and animal sciences, economics, social sciences, medicine, and more). Using lectures, discussions, roleplays, and lots of hands-on work with OSG experts in HTC, participants will learn how HTC systems work, how to run and manage many jobs and huge datasets to implement a full scientific computing workflow, and where to turn for help and more info.

 Worried about costs? Successful applicants will receive financial support to attend the OSG School, covering all basic travel, hotel, and food costs. This is a great deal! Ideal candidates are graduate students whose research involves or could involve large-scale computing - work that cannot be done on one laptop or a handful of computers. And every year, we accept some post-doctoral students, faculty, staff, and advanced undergraduates, so make a good case for yourself!

IMPORTANT DATES

Application Period (OPEN NOW): 14 March - 15 April 2016 OSG
User School: 25 - 29 July 2016

MORE INFORMATION AND APPLICATIONS
Web: http://www.opensciencegrid.org/UserSchool
Email: user-school@opensciencegrid.org
Facebook: https://www.facebook.com/OSGUserSchool
Twitter: https://twitter.com/OSGUserSchool

Announcing OSG Software versions 3.3.11 and 3.2.37

We are pleased to announce OSG Software versions 3.3.11 and 3.2.37.

Both 3.3.11 and 3.2.37 include:
* VO Package v65 - more OSG CA transitions
* CA certificates based on IGTF 1.73

Changes to OSG 3.3.11 include:
* XRootD 4.3.0: Several important fixes for bugs affecting CMS
* HDFS 2.0.0+1612: Support ACLs, Support the EL7 platform
* Update to GlideinWMS 3.2.13
* Add gfal functionality to xrootd-dsi
* HTCondor CE 2.0.4: Accept full subject DNs in extattr_table.txt
* BLAHP 1.18.18: Changes in the BLAHP to support PBS Pro
* osg-pki-tools 1.2.15: Better error messages and checking of arguments
* HTCondor 8.4.5: Various bug fixes
* Pull in the required log4j package when installing the emi-trustmanager
* HTCondor 8.5.3 in the Upcoming repository
* Support for an OSG CVMFS configuration repository in the Upcoming repository

Release notes and pointers to more documentation can be found at:

https://www.opensciencegrid.org/bin/view/Documentation/Release3/Release3237
https://www.opensciencegrid.org/bin/view/Documentation/Release3/Release3311

Need help? Let us know:

https://www.opensciencegrid.org/bin/view/Documentation/Release3/HelpProcedure

We welcome feedback on this release!

Tuesday, April 5, 2016

GOC Service Update - Tuesday, April 12, 2016 at 14:00 UTC

The GOC will upgrade the following services beginning Tuesday, April 12 at 14:00 UTC. The GOC reserves 8 hours in the unlikely event unexpected problems are encountered.

OIM

  • Bugfix for previous release to allow alternate names for host certs in different domains https://jira.opensciencegrid.org/browse/OIM-135, code release 3.51
  • Adding signer search for user certs
  • Change to remove renew/request option for ATLAS