After about 36 hours the DAQ had 273 defunct processes and was failing to read most of the time. Stopped and restarted it. Minor code cleanup and debugging added to MLP.cc to help trace that persistent fake error message.
Failures reappeared after 6 cycles! 78 defunct processes so far.
Terminal server 01 is offline again. That probably accounts for the sudden onset of failures at about 6pm CERN time. No, it doesn't. The terminal server seems have been off from the getgo.
I created a new monitoring script ~emudcops/DCOPS_READOUT/monitorTS.sh
It writes a new line into ~emudcops/DCOPS_READOUT/TSBAD every time it finds a terminal server offline (and it checks every 20 minutes). If the PC reboots we should restart this script.
Update: 0T rather than 3T!
Looking at the raw 0T data from 29-Aug and eyeballing the profiles (junking ones that I consider bad--including a couple that would fit, but with rubbish), I calculated a few numbers. (A couple of DCOPS could not be contacted.)
Bottom line: We can find coordinates for 88% of our DCOPS CCD pairs.
How many DCOPS are currently not useable in any direction? Of 504 total DCOPS/laser combinations:
Bottom line: 3.8% of the DCOPS/laser combinations currently lack any useful data in any direction.
For many of these (esp. transfer lines) this can and will be fixed. 12 of the no-data DCOPS are on the transfer lines, and the lasers are not yet in final position. (The 2 DCOPS that don't answer are also on the transfer lines: 4 and 5 at position 5 {0=+4, 11=-4})
Modified 11-September-2008 at 09:14
http://hep.physics.wisc.edu/~jnb/cms/11Sep2008
| Previous notes | Next notes | Main slide directory |