As far as CODA stalls go, tests I did last week show one of the ROCs
will sleep at data rates over 60kB/s. The problem seems to be the data
buffers in the ROC (500 events' worth) are full, and the EB (event
builder) is unable to accept data, so the ROC won't accept any more
triggers. At some point the logjam breaks, and the ROC takes data again
for a while. The "up" time seems to average around 100 seconds, the
"sleep" time is also around 100 seconds.
But there can be large variations in both times.
I'm still investigating exactly what causes the logjam, and how to free
it.
Karen
Chris Crawford wrote:
>
> better late than never ...
>
> Saturday 16:00-24:00
>
> * we took an hour of data with in 'no strobe' mode, 'ts2_pri_ns.crl'
> * then we tried a new elastic trigger, 'elastic_tof10-15.settings'
> with all 0's replaced by x's in the mlu TOF bits (to allow more
> than one TOF being hit, back to 'common strobe' mode
> - there were some problems with beam quality, and so they adjusted
> some of the quad settings, and opened the slits a little
> - after that, the beam had a very nice lifetime (10 min)
> and the current was very smooth as a function of time
> * we finished by taking an hour of data with no B-field
> - coda was stalling frequently even at data rates < 300kB/s
>
> --Genya, Doug, Chris
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:28 EST