Using Evidence to Diagnose Problems

Understanding Exit Codes Signals

When a Broker or agent is terminated, the exit code can be seen in the log file (for example oplrqb.log) or in the foreground window.

The exits codes are shown below:

Table: 1. Log Levels

Exit Code	Category	Comment
0	NORMAL	Normal exit with success
1	WARNING	some kind of failure at startup/shutdown, typically shown with a reason for this error
>1	UNKNOWN	When the exit code is not in the range of 0-15 then it is typically an uninitialised exit value taken from the stack. This can occur when an error in some database code is returned to the OpenLink code (Broker).

The agent or broker may be terminated with a signal:

Table: 2. Reported Signals

Signal	Name	Category	Comment
2	SIGINT	NORMAL	user pressed CTRL-C on oplrqb -fd
3	SIGQUIT	NORMAL	user pressed CTRL-\ on oplrqb -fd
4	SIGILL	FAULT	illegal instruction resulting in a program crash
7	SIGBUS	FAULT	misaligned mem read/write resulting in a program crash
8	SIGFPE	FAULT	divide by 0 or other arithmetical error
9	SIGKILL	NORMAL	sent by the broker to kill an agent which has ignored multiple SIGTERMs
11	SIGSEGV	FAULT	buffer/stack overrun resulting in a program crash
13	SIGPIPE	NORMAL	other side of a pipe was killed while this side tried to write data to it
15	SIGTERM	NORMAL	sent by the broker to an agent to ask it to terminate

On the Windows platform, severe errors are reported in the following way:

pid ... died with status code 0x....

The hex number may be looked up in the winnt.h file. All errors reported in this way are considered severe, and should be reported to OpenLink Support.