Presentation is loading. Please wait.

Presentation is loading. Please wait.

Jan 2010 OSG Update Grid Deployment Board, Feb 10 th 2010 Now having daily attendance at the WLCG daily operations meeting. Helping in ensuring tickets.

Similar presentations


Presentation on theme: "Jan 2010 OSG Update Grid Deployment Board, Feb 10 th 2010 Now having daily attendance at the WLCG daily operations meeting. Helping in ensuring tickets."— Presentation transcript:

1 Jan 2010 OSG Update Grid Deployment Board, Feb 10 th 2010 Now having daily attendance at the WLCG daily operations meeting. Helping in ensuring tickets are handled timely and interface issues identified more quickly. The plan is to cover these with someone even if the operations team is unavailable for some reason. Production Incidents Update:  A bug was fixed in the BDII at CERN that resulted in stale entries. The GOC's monitoring of the BDIIs has demonstrated that everything is working appropriately.  A transition was made to a web services based system that interfaces directly between the GOC and GGUS, should help improve the reliability of ticket exchange.However in the short term it resulted in Alarm tickets over the weekend not being delivered to US ATLAS BNL site. We are addressing these issues. NSF OCI calls for SDCI and STCI prompting discussion of what OSG Satellite proposals will be beneficial, opportunity for funding for continued support of middleware (Globus, MyOSG) and Build and Test. OSG All Hands Meeting March 8-11 th. We apologised for the overlap with the Taiwan ISGC 2010. We will remember this when scheduling in the future. 1

2 Jan 2010 Installed Capacity. Installed Capacity reporting ready and waiting for test reports back for validation. Remember that OSG provides a central service for this where US LHC management can validate and correct (e.g. publish the pledge even if the actual is higher. ) Will work with Laurence and John on difference between OSG format and gstat2 for validation. Example table below. 2 SiteSI2KHepSpec06TBInstalled MWT2_UC1000400300 MWT2_IU1500500200 Total: US-MWT22500900500 Total: USATLAS

3 Jan 2010 Testing new CE Services – CREAM and GT5 We have been evaluating Cream and GT-5 for more than 6 months. For GT-5 the Globus team has been providing a lot of necessary support to sort out issues of performance, integration and instability. Results of the tests are at Rates of up to ~2Hz (100 jobs/min) were observed. Use condor-G 7.4+ with its gt5 mode. For Cream we can now integrate with Condor as the local batch system and Condor-G as the client. US ATLAS and US CMS have single installations for testing. We are writing a list of all tasks and decisions needed to deploy each of these into production on the OSG. We will review this list in mid-March. 3

4 Jan 2010 Tier-3 Support Documentation written as a hands-on guide to enable US ATLAS and US CMS Tier-3 system administrators to deploy a "new" cluster. The basic cluster installation creates a baseline infrastructure that experiment specific application components can be layered on top of to complete the intended functionality of the Tier-3 system. It includes a “security checklist” and configuration of Xrootd and Bestman storage interfaces. https://twiki.grid.iu.edu/bin/view/Tier3/WebHome https://twiki.grid.iu.edu/bin/view/Tier3/WebHome US CMS used documents for recent UCLA & Purdue Calumet installations; Expect new Tier-3s to be purchased in February-March. Tier-3 liaisons continuing to investigate VMs as a way to reduce site- administrator loads in support of high level services. Starting to investigate use/support for higher level services – e.g. US CMS Tier-3-wide Phedex. 4

5 Jan 2010 Storage & Data Management Increased requests from non-LHC VOs for use, reservation, sharing of storage and data management tools. Requests quite diverse.  Earthquake engineering simulations need 3 TB of storage space for runs of 10,000 jobs each of 30 minutes duration.  Fly’s Eye needs 5 TB/site with >4 sites for more than 1 year.  D0 needs 500 GB for runs of 9 hours.  IceCube needs ~15 GB persistent at site ~1GB per WN. Not clear the best way to use the existing experiment specific storage and SRM required interfaces to meet the needs. Considering “storage depots” provided by OSG for sites to install for public storage. 5

6 Jan 2010 Security Debian CA repository being supported for LIGO. Working with David Group to improve fetch-crl error reporting in particular OSG configuration failures. Evaluating/deploying pakiti tool for the US LHC Tier-3s. Working with Romain on fixes/extensions. Request to OSG for improved local documentation. Changes in Openssl being evaluated for impact on existing scripts and code. Expect some changes needed. 6


Download ppt "Jan 2010 OSG Update Grid Deployment Board, Feb 10 th 2010 Now having daily attendance at the WLCG daily operations meeting. Helping in ensuring tickets."

Similar presentations


Ads by Google