Description of problem: At Cisco we have a wrapper shell script to execute rhnreg_ks and if successful (returning 0), removes a symbolic link in the system's /etc/cron.daily/ directory. We're encountering cases when it appears to have succeeded in registration but it didn't return the right error code, so the symlink wasn't removed. Then later, we get a legit error because the machine is already registered under those credentials. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. rhnreg_ks Additional info:
what are the cases where it returns the wrong error code, and what error code does it return?
We were hoping you could tell us that. There is no stderr when it successfully registers and doesn't return the right error code, so cron reports nothing. It's happening about 1 out of 20 times.
can you please run that script through some sort of enhanced set -x so we know what is going on - you're asking us to shoot in the dark here, with nothing more than "sometimes there is an error" to start with.
Any updates on this? Have you been able do duplicate the problem, or set up some kind of capture script?
Sorry for the lateness; I was out of town for a few weeks. As I stated earlier, it's sporadic at best. The only time I see the problem is after it happens, on the *next* day's cron output, which I get from every single Linux box at Cisco. I'm not going to run all of them through -x for something that happens so sporadically. If I see it happening more frequently I'll consider it more seriously.
No action for 6+ weeks, unable to isolate or duplicate, so closing.