Reanalysing a run Online, e.g., when recovering from the "dropped datagram problem"

Determine *when* the runs were taken. You'll need the day number of the year. One method is:
```
python
>>> from datetime import *
>>> datetime(<yyyy>, <mm>, <dd>).toordinal() - datetime(<yyyy>, 1, 1).toordinal() + 1
```
Copy the data from /gnfs/data/DVD/RawPackets/2006<day of year>/ to a local directory, say ~/sci. (It is probably best to use the glast account for this to ensure that you are working with the correct environment.) The date and time found in the eLog runs section is the same as used in the .../RawArchive/* directories. Only the *_SCI.pkt files that span the run of interest are needed.
mkdir ~/sci/lsf
Using the glast account on a lat-licos* or equivalent (i.e., a machine that has LICOS_Scripts installed):
```
python $ONLINE_ROOT/LICOS_Scripts/analysis/LsfWriter.py --lsd_dir=~/sci --lsf_dir=~/sci/lsf
```
Determine which of the output files is the run you want. File format is:
```
<ApId>-<GroundId>-<Time>-<nFirst>-<nEnd>.lsf
```
GroundId is the run number, without the 077000000, in hex.

Verify that nFirst is 0000.

Login to lat-dmz01 and enter the following to connect to the MySQL database:
```
mysql --user elogbook elogbook --password
```
(enter the password at the prompt)

Enter the following command to reset the analysis state:

update LICOS_activities set JobCompleteList = '[None]', AnalysisState_FK = 7 where RunID_FK = '999999999';

(replace 999999999 with the acquire run id)

Copy the lsf file(s) to /nfs/data/lsf. If the corresponding DAQ run had multiple steps, there will be multiple lsf files.
AnalysisEngine will see them and should be able to reprocess the run.