Log entry time 17:42:51 on March 2,2010

Entry number 310467

17:18 The timing board is in a normal state (fig 1).

Do another round of DAQ - 240 Hz - ET tests, but with a small change
to THaEtClient to measure the rate over 10 sec instead of 5.

Run 50810

Offline Pan for 10K events:
Events failing cut conditions:
0  Low_beam        10000 
1  Beam_burp       0     
2  Evt_seq         0     <-- low, hence low deadtime
3  Pair_seq        4     <-- low, ditto
4  Startup         1     
5  Mon_Saturate    0     
6  Adcx_DAC_burp   2     
7  Adcx_Bad        10000 (clearly a database problem, standalone 
code sees not so many "bad")

Looking at panFFB log file:
Asymptotic region --

ET rate 195.1 Hz in 10 sec, avg 198.2 Hz
ET rate 190.1 Hz in 10 sec, avg 198.1 Hz
ET rate 205.1 Hz in 10 sec, avg 198.2 Hz
ET rate 210.1 Hz in 10 sec, avg 198.4 Hz
ET rate 195.1 Hz in 10 sec, avg 198.3 Hz
ET rate 200.1 Hz in 10 sec, avg 198.4 Hz
ET rate 200.1 Hz in 10 sec, avg 198.4 Hz
ET rate 195.1 Hz in 10 sec, avg 198.3 Hz
ET rate 190.1 Hz in 10 sec, avg 198.2 Hz
ET rate 210.0 Hz in 10 sec, avg 198.4 Hz
ET rate 200.1 Hz in 10 sec, avg 198.4 Hz
ET rate 210.1 Hz in 10 sec, avg 198.6 Hz
ET rate 200.1 Hz in 10 sec, avg 198.6 Hz
ET rate 210.1 Hz in 10 sec, avg 198.8 Hz
ET rate 215.1 Hz in 10 sec, avg 199.0 Hz
ET rate 205.0 Hz in 10 sec, avg 199.1 Hz
ET rate 210.1 Hz in 10 sec, avg 199.3 Hz
ET rate 195.1 Hz in 10 sec, avg 199.2 Hz
ET rate 210.1 Hz in 10 sec, avg 199.3 Hz
ET rate 195.0 Hz in 10 sec, avg 199.3 Hz
ET rate 210.1 Hz in 10 sec, avg 199.4 Hz

This suggest a rate of 200 Hz, but I think it is artificially low.
A more reliable measurement (probably) is from the cuts summary:

tail -1000 ffb_50810.log

Cumulative stats for 267044 events: 

Events failing cut conditions:
0         Low_beam      2
1        Beam_burp  15785
2          Evt_seq    787  <-- 0.3 %
3         Pair_seq   4393  <-- 1.6 %
4          Startup      1
5       Low_beam_c      0

Tentative conclusions:
^^^^^^^^^^^^^^^^^^^^^^
1. Running on adaql5 causes no deadtime (see offline pan result).
This is expected: Gigabit ethernet, fast disks, etc.

2. Running ET on adaql5 causes a 2% loss of data, good enough
for feedback.