NAT - segmentation fault

, ,
1

Hi, here is the situation.

I am still using NAT611 on AIX, and I have been receiving an error after executing Natural Batch. The scenario is a little bit complex, but let me try to explain it.

Scenario: Session A, on machine A, is started a Natural online process. This process, after receiving some information from the user, it starts a SUDO command to execute a script via remote shell to the machine B, and this script starts (on machine B) a Natural Batch process as Session B. This session B starts another Natural Batch session - C, at the same machine B.

Problem: Depending on the options using application on session A, the session B works fine, however, using almost all the options in the application on session A, I have been receiving a segmentation fault error - Signal 11, system code 50, on session B - machine B (remember, it was executing by remote shell via sudo)

It seems that when the session A has a few statements to be executed, it works, but if there are several statements, it fault.

I am not an Unix expert, there may be some parameters on Unix that I dont know if it is OK or not. For example: the limits on AIX is OK (/etc/security/limits), how do I know if it is OK, at this moment, the default value on limit file is -1, and I assume it means unlimited, and it does not matter how many resource I need, it could use at the top. SEMMNI, SEMMNS, etc…how it would be the best choice, is there a magic number? is it related to this problem?

It also seems that if we try to reduce the size of the object/application, there is a chance that it works, but I do not know if there is something more that we can do before changing the application code.

Some doubts, as part of the abend message, it says something about the thread,I am not sure if it is possible to increase the thread size, in order to be able to execute the entire statements (I also dont know if there is this concept on Unix).

There is no problem with memory on this machine, neither with the processor.

As I said, we would like to solve the problem, first, changing some parameters, if it is possible, and it could be Natural parameters, such as USIZE (200MB right now), …, or operating system parameters. So, if someone has any idea or suggestion, please, let me know as soon as possible.

tks.

2

Are there any errors in the Natural output? Are you simulating batch or running batch mode?

3

Hi James,

there is no Natural message error, and, yes, the abend occurs on Natural Batch, however, you need to keep in mind that the Natural Batch is started after some other processes. Look at the scenario, Session A, machine A (Natural Online) → (shcmd) SUDO → starts a remote shell process → Start Session B on machine B (Natural Batch) → …

I was thinking that there could have some AIX parameter that it would be necessary to change, such as SEMMNS, SEMMNI, …but, as it is an AIX machine, it seems that the parameters dont need to be set, it is auto adjusted. I am not sure about it, but if anyone could say something different, please, let me know.

Before anyone ask me, I think it is not a Buffer Pool problem, because the statistic show me that there are several bytes free, and besides this, there is no buffer pool message error as well.

tks