I've been looking into migrating from 3.6 to 4.x on and off since 4.0 first came out. I'm pretty close at this point. I managed to migrate our custom field data, but it was a quick and dirty method. over the years (we started using Cerberus at v.1.3) we had various custom fields, some of which we no longer care about, so I was only migrating fields who's field ID was > 10 - I also created a bunch of contact custom fields we're now importing from Active Directory (via a CGI script which uses LDAP and exports CSV), so I don't need/want to migrate those fields. I also had to decide if the new custom fields would be Global or group based. I made them global for our case, but perhaps I'll rethink that and make them group. Logically for us, they should probably be bucket-based, but that's not available at this version, so...

I did make sure to create the new custom fields in the same order as the old ones by ID number, so all I had to do was subtract 2 from the old field_id value to map to the new custom fields' IDs. If you want to try this script, you'll have to adjust the "WHERE v.field_id > 10" clause, and the "SELECT v.field_id - 2 AS field_id" clauses to match your mapping requirements, but here it is in a few variables, a couple password prompts, and one long 4-pipe shell command:

Another warning - I ended up combing through the schema in 4 to find which tables to delete all rows from, and which sequences to reset so I could re-import everything without having to set up all the agent accounts, custom fields, groups, etc. - and I've had to do so often while developing this script. I've also cleaned out soem site-specific usernames and such and have not tested it since I did so. Caveat Emptor

Oh yeah, all my custom fields were stringvalue ones - this gets harder if they have different types as well. I'm using a temp table, because I couldn't figure out another way to map the old ticket id's through the masks to the new ticket ID's in one pass, since the ID's only exist in their own databases.

#------------- cut here ------------------

echo "Migrating custom field data from cerb3 instance to cerb4 instance"
echo "Assuming field_id from cerb3 is just 2 higher than corresponding "
echo "field_ids in cerb4. If not, adjust script for your environment "
echo

# hosts to connect to via mysql command line utility. include -h, or
# leave blank for localhost

CERB3HOST="-h cerb3server"
CERB4HOST=

CERB3USER=cerb3dbuser
CERB4USER=cerb4dbuser

CERB3DB=cerb3
CERB4DB=cerb4

TEMPHN=${CERB3HOST?localhost}
CERB3HPROMPT=${TEMPHN#-h}

TEMPHN=${CERB4HOST?localhost}
CERB4HPROMPT=${TEMPHN#-h}

read -s -p "MySQL password for $CERB3USER on $CERB3HPROMPT: " T3PWD
echo
read -s -p "MySQL password for $CERB4USER on $CERB4HPROMPT: " T4PWD
echo

echo 'SELECT v.field_id - 2 AS field_id,
             t.ticket_mask,
             v.field_value,
             "cerberusweb.fields.source.ticket" AS source_extension
      FROM field_group_values as v JOIN ticket as t ON (v.entity_index = t.ticket_id)
      WHERE v.field_id > 10
      ORDER BY v.entity_index;' \
 | mysql $CERB3HOST -u $CERB3USER --password=$T3PWD $CERB3DB \
 | awk -F'\t' 'BEGIN {print "DROP TABLE IF EXISTS temp_stringvalue;";
                      print "CREATE TEMPORARY TABLE temp_stringvalue (field_id INT(11), mask VARCHAR(30), field_value VARCHAR(255), source_extension VARCHAR(255));";
                     }
                   {print "insert into temp_stringvalue values (" $1 ", \""$2"\", \"" $3 "\", \"" $4 "\");"}
               END {print "INSERT INTO custom_field_stringvalue \
                           SELECT v.field_id, t.id, v.field_value, v.source_extension \
                             FROM temp_stringvalue AS v \
                             JOIN ticket AS t \
                               ON (v.mask = t.mask);"}' \
 | mysql $CERB4HOST -u $CERB4USER --password=$T4PWD $CERB4DB

#----------------------- cut here -------------------