One of my customer is running Tuxedo application server on Solaris SPARC 10 and they found that application server process crash every day around same time but it not generating core, without core debugging this crash is going to be difficult. There are following conditions where Tuxedo application process crash but don’t generate core:
a) If exit() or _exit() function called inside server code. Customer verified they have not used it.
b) If application catch the signal e.g. SIGSEGV,SIGBUS. Then application will be terminated by admin process and core will NOT be generated.
Since this application process is getting crash everyday and almost around same tine I can ask them to enable trace, truss is going to be very heavy as it is on production, so I thought dtrace would be ideal. But I am new to this tool, can you please guide how I should use this to get right information.
Please note Tuxedo application process is c or c++ generated binary, and coreadm looks alright.
global core file pattern: /data3/var/core/core_%n_%f_%u_%g_%t_%p
global core file content: default
init core file pattern: core
init core file content: default
global core dumps: enabled
per-process core dumps: enabled
global setid core dumps: enabled
per-process setid core dumps: disabled
global core dump logging: enable