This describes the steps I took to create miniminidvodbs (and smaller) to try to find the corrupt SMF.

Step 1:  Find the cam_ids used in !ThreePi.283.  You can find them with this mysql query:
{{{
select cam_id from camRun join camProcessedExp using (cam_id) join addRun using (cam_id) join minidvodbRun using (minidvodb_name) where minidvodb_name like '%283';
}}}


Then, take the first half and the last half and create 2 minidvodbs, using one addtool command for each cam_id. This shows 2 examples:

{{{
addtool -definebyquery -destreaked -set_dvodb DEBUG.THREEPI.283.1 -set_minidvodb_group debug.ThreePi.283.1 -set_label debug.ThreePi.283.1 -dbname gpc1 -cam_id 148883
addtool -definebyquery -destreaked -set_dvodb DEBUG.THREEPI.283.2 -set_minidvodb_group debug.ThreePi.283.2 -set_label debug.ThreePi.283.2 -dbname gpc1 -cam_id 149485
}}}

Edit site.config in ~ipp/ippconfig to know about DEBUG.THREEPI.283.1 and DEBUG.THREEPI.283.2

in addstars pantask:

{{{
addstar.off
minidvodb.merge.off
add.minidvodb Debug.ThreePi.283.1 DEBUG.THREEPI.283.1 DEBUG.THREEPI.283.1 1 GPC1
add.minidvodb Debug.ThreePi.283.2 DEBUG.THREEPI.283.2 DEBUG.THREEPI.283.2 1 GPC1
add.label Debug.ThreePi.283.1
add.label Debug.ThreePi.283.2
(remove all the old labels)
}}}

Wait until there are entries for !Debug.!ThreePi.% in the minidvodbRun table. When there are, type this in addstars pantasks:
{{{
addstar.on
}}}


Now run addstar and relphot:

{{{
addstar -resort -D CAMERA GPC1 -D CATDIR /data/ipp004.0/gpc1/catdirs/Debug.ThreePi.283.1/
relphot -averages -update -D CAMERA gpc1 -D CATDIR /data/ipp004.0/gpc1/catdirs/Debug.ThreePi.283.1
}}}

Results:

 * !ThreePi.283 - segfault
  * !ThreePi.283.1 (first half of !ThreePi.283) - completed succesfully
  * !ThreePi.283.2 (second half of !ThreePi.283) - segfault
   * !ThreePi.283.4 (second half of !ThreePi.283.2) - completed successfully
   * !ThreePi.283.3 (first half of !ThreePi.283.2) - segfault
    * !ThreePi.283.5 (half of !ThreePi.283.3) - completed successfully
    * !ThreePi.283.6 (half of !ThreePi.283.3) - completed successfully
     * !ThreePi.283.7 (half of !ThreePi.283.6, before I found out it didn't segfault)
     * !ThreePi.283.8 (half of !ThreePi.283.6, before I found out it didn't segfault)

So, chopping in half didn't work.  !ThreePi.283.3 segfaults, but it's 2 halves do not.

Current attempts:

 * Taking a copy of !ThreePi.283.6, and do the following:
   1. add a cam_id from !ThreePi.283.5 (6 and 5 make up the 2 halves of !ThreePi.283.3)
   2. run relphot on !ThreePi.283.6. Does it segfault. If not, go back to step 1.

 * recreating !ThreePi.283.3 using the ipp-20101029 tag. Does it segfault with relphot?