1) Progress and plans: - Florida Globus 2.0 beta installed (not yet configured): The binary install using the new Globus Packaging Tools (GPT) was very easy and worked the first time! The next step is to configure. Rick asked if he could just use the gatekeeper certificate from an earlier globus 1.1.4 install. Jim thought yes, but had not actually tried it. GDMP 2.0 attempted installation: The head node of the Florida cluster has rpm version 3.0.4. Rick reported that he gets back an error message indicating that rpmlib versions <= 3.0.3-1 are needed by gdmp. This seems strange from a backward compability perspective. Jim pointed out that rpm install of gdmp may (or may not) need to have globus installed via rpm. Jim offered to check. In addition, Conrad also pointed out that rpm under went a significant change between 3.0.3 and 3.0.4... Condor 6.3.1 installed (and checked) This is from an already existing installation. - Wisconsin Ordered new HW for the cluster. The vender ran out of supplies for the video card and delayed the delivery of the machines. Good news: the machines were shipped on Wednesday and are expected to arrive any day... - San Diego Ian reports that San Diego is in good shape: the HW is fully set up and include 3 machines in a rack. They are fully configured and talking to each other. The next step is to install the VDT software. - Caltech Suresh reports that they have recieved the 1st front end processor pre-installed with RH 7.2, including a gigabit card. At the moment, work is progressing to down grade from the pre-installed RH 7.2 to RH 6.2. Also a 5 gigabit switch has been received, and Caltech has made a request for a fibre connection between seperated two Caltech sites. It is hoped to use this connection on the US-CMS Test Grid for testing networking issues. - Fermilab Globus 2.0 beta installed (not yet configured) on a test machine: Jim reports that the binary installation of Globus 2 using GPT went smoothly with the exception of the /opt/globus2/bin directory, which was created as a file rather than as a directory. This was fixed by creating the directory by hand and then re-installing. Florida did not see this problem. The knowledge used here will be used to install Globus on the Test Grid machines (see below). 3 machines (Pentium 4 1.5 GHz) will arrive for use in the CMS Test Grud by the end of the month (but probably won't be configured until after GlobusFest '02.) Greg has asked Natalia and Yujun to set up a web area containing all available DAR distributions. This should be set up soon. (The rigamarole is to get enough afs disk space where we can serve the files.) This will be password protected however because of licensing issues. I will update you with a password when ready. 2) Testing and commissioning: Florida, Fermi, and San Diego expect to have at least a subset (including Globus 2.0) of the VDT 1.0 installed before the end of next week. Hence, it was decided that a simple "Hello World" as well as job submission tests will be conducted between UF, FNAL, and UCSD before 25 Jan. Ian, Jim and Rick will communicate via email to accomplish this. This will create a on-line environment for Wisconsin and Caltech when they are ready to test. 3) Future Plans - US-CMS/GriPhyN/PPDG Virtual Data Grid prototype (proposed by Koen) Rick pointed out that there was an earlier telecon on Wednesday (not to be confused with this test-bed telecon) which discussed a prototype Virtual Data Grid architecture for CMS. This prototype is an attempt to Gridify the CMS production environment using existing tools wherever possible. In particlar it will use: IMPALA, BOSS, a Virtual Data System (V1.0), MOP, GDMP (replica cat), etc. Development and testing are foreseen to take place on the US-CMS Test Grid during late Spring and early Summer. The aim is to have a working prototype for Autumn. - Any hardware experimentation? (Not discussed in the meeting. To be discussed next time.) - Coordination with Claudio Grandi and the EU-CMS Test Grid? (not discussed in the meeting) Rick is in contact with Claudio regarding the integration of the US-CMS Test Grid (based on VDT 1.0) with the EU-CMS Test Grid (based on EDG middleware) but, so far, no immediate/concrete plans. Rick will report about progress in the next meeting. - CA/RA issues? Conrad reported about his study of the EDG CA. He find it to be well thought out along with a nice GUI interface. His investigation is still in the early stages, but will likely to testing on the US-CMS Test Grid. - Authorization issues? (not discussed in the meeting) Rick spoke with Ruth about authorization issues on Thursday. It seems that Globus will make CAS (Community Authorization Service) available soon to a "select" group of users. No word yet when CAS will be alpha, much less beta. But, we might have some interest in asking Globus if we could be included as on of their "select" preview groups. 4) Any unfinished/important conclusions from the last telecon. Harvey pointed out that the US-CMS Test Grid is valuable for studying hardware and networking issues. Hence, the Test Grid should be used for hardware/networking experimentation as well as software development. Everyone agreed that this is very important and should be one of the primary uses of the US-CMS Test Grid.