Level 3 Online Status Reports

Back to index Previous Next

Issue 5
Monday, April 26, 1999

Event Flow

Reports

• An L3->CS interface has been specified: the new event record will consist of the usual 4-word message header (which will have network byte order when sent), 32 4-word spaces, followed by the event data itself. These 32 words should provide stability against future changes; in the meantime, the Level 3 executable can use it as it wants to provide routing or monitoring information or pointers to such information. These words also allow other test clients to include their own blocks of additional data outside the event data record itself.
• Test wedge and TDC's were only sporadically available last week due to maintenance work, upgrades, cooling shutdowns, power outages...

Actions

• Jeff will implement the L3->CS interface on the L3 side.
• The SCPU will be updated for 6-word commands.
• Ilya will resume wedge tests when the test wedge is available again, and build a larger system test.

Executable Interface

Reports

• Christoph has written a Java GUI for the raw data generator. This will write a specification file which is read by a C program (written by Andreas) which uses Christoph's data generation library. The C program then generates the data and puts them into a file.

Actions

• Christoph will complete the data generator GUI.
• Andreas and Christoph will test drive it.

Test Control

Reports

• Steve has written an IDL conventions guide, which can be found here.
• Using the scanner manager IDL, Ilya and Sasha have operated an ILU C server and a Java ORBacus client. b0l3pcom1 has been used as the server, and both b0l3pcom1 and gobi have been used as clients.

Actions

• Ilya will implement user-defined exceptions with C binding in ILU.
• Sasha and Steve will extend the functionality of this program.

Monitoring

Reports

• Mike has revised his ROOT I/O code to use the TMapFile class.
• Andreas, Ivan, and Jeff have continued discussions on DIM and its shortcomings, but whether or not it can be simply modified is unclear.

Actions

• More investigations on DIM.
• Jeff will add status information to l3_converter.

Physical

Reports

• FNAL VxWorks PPC kit is now available.
• Purchase orders are out for major components. Cables and packaging haven't been done.
• Steve reports on the switch control software: Fore 5.2 is the newest software that can be used on the old SCP; it is also the oldest version that can be used on the new SCP. It is also a free download (version 6 requires an upgrade fee).
• Fore has an engineer visiting Argonne every month; if we tap into him, he might do the switch fabric installation gratis. Otherwise there is an ~8% installation fee.
• Rob Roser has ordered an "evaluation shelf." The footprint is 2x4 feet (1x2 panels). He says that the technicians can do the cabling for us, but we have to tell them what to do.
• Jeff sketched a first-pass layout for the 3rd floor computing room. The network switching is left in the center, and the CS/L and database servers are nearby. Two feet of space is left between each row of shelves (each PC has about 1.5 foot depth). This gives us capacity for 24 shelves. Assuming 4 shelves per unit and 5 PC's per shelf, this means 480 PC's.

Actions

• Steve will reclaim VME CPU's from CD and send them out for upgrade.
• Jeff will get information on cabling and trunk cabling options, and provide an initial design of the network layout.

Back to index Previous Next