Hello Mr Rude,

Well, it's a little bit strange. A link between dmfcrp (recovery) and gcc, I
do not think there is any link between recovery and Ingres/NET ... but I
don't know all the internals.

Logs files are (in $II_SYSTEM/ingres/files) :
- errlog.log
- iiacp.log (or II_ACP.LOG) : archiver log (dmfacp), responsible of
journaling process
- iircp.log (or II_RCP.LOG) : recovery process (dmfrcp)

You can also set this variables :
ingsetenv II_DBMS_LOG $II_SYSTEM/ingres/files/dbms.%p.log
- %p is mandatory to have separate logs for differents processes and since
OpenINGRES (if my remember is correct) dmfrcp is just an other iidbms. %p
mean Unix PID
- restart Ingres after that

If you say reboot correct your problem I think it could something like a bad
file allocation (system) or some system tables not correctly synchronised,
or full, or something like that. I know AIX is not like all others Unixes in
some situation cause they are dealing with their own DB and sometimes it can
give some issues (problem to gain a process number in some huge situation,
etc).

Is there is something relevant in syslog (or errpt -a if my remember is
correct) ?

Try to look in the servicedesk.ingres.com knowledge database and raise an
issue to Ingres Corp. There are many many situations in their KB and as far
as I know many support and developpers team are the same since the many
years, so they have a great and complete experience and knowledge of many
situations ...

Hope this help,