so the problem is that server crashes or that after a crash tuxedo is unable
to restart it or both? are you able to start it manualy?
check if server dumps a core (inside: "tmunloadcf|grep APPDIR" directory). if yes
then it might be helpful to server author
check if server produces its own log file. perhaps it has ability to create one
if configured properly?
if "tpsvrinit() failed" then you might want to analyze server boot with system
tools (truss on aix, strace/ltrace on linux)
if server is in resource manager group then check if this start problem
is caused by database/mq/whatever connect problem
contact server author
"error: 23" is an OS error. It is platform dependent but in general I think the errorno is defined as:
#define ENFILE 23 /* File table overflow */
You probably ran out of OS resources when tpsvrinit() tried to do a file open (e.g. maybe initializing a DB connection).
Either the OS kernel is not configured properly for the application architecture requirements or the resources are not being reclaimed properly.
Since this is occurring after a server crash I would think it was the second problem. Clean up resources (e.g. ipc, dead processes).
Figure out why the server crashed. Make sure you have the highest Tuxedo 9.1 rolling patch level available installed.