With more statistics I can say that some DCOPS fail more often than
others, but not much. This is a display of DCOPS board ID for the Plus side
for failed reads.
One failure occurs at line 724, not at the beginning of the data block.
Essentially all look like line-noise corruption, not belated commands.
I have 29 failures in 445 events (180 reads per) for a 0.036% failure rate.
Updated 6-Aug
The program crashed after 5-Aug 2:33. Something seems to have failed; possibly the low voltage. Afterwards the program tried to connect to ports which hadn't released and crashed. It restarted, of course, but in the usual (correct) STANDBY state.
Update 7-August. I restarted the program, but with both Plus and Minus enabled. I'd checked that Minus was working. Apparently between the time I checked and program run was issued the LV went off. The program crashed quickly, and I found many port already in use errors.
I had to reboot alignterminal01, after which it worked again. This may get to be a problem, and require some automation.
Modified 07-August-2008 at 11:35
http://hep.physics.wisc.edu/~jnb/cms/04Aug2008
| Previous notes | Next notes | Main slide directory |