Last week I ran into a situation where, after having executed the AI server-side test suite, I ran into a problem with doing actual installs on x86. After the suite had completed (and verifying that all AI services were off) I configured the x86 AI services on the server and tried to net boot an x86 client off of it. The client would get a dhcp address but time out at the tftp step. Rebooting the server and trying again cleared the problem.
Today I've run into the same problem again using the 111b2 packages. However, this time, rebooting the server doesn't help. The x86 client still gets stick at the tftp stage. I've tried this with both x86 and sparc machines as the AI server and get the same result. Has anyone else run into a problem like this? The AI server-side test suite does a lot of creation and deletion of services, and goes to great pains to try to make sure previous configurations are cleaned up (no services listed by 'installadm list', no AI related filesystems mounted, all associated httpd and webserver processes killed off). I'm wondering if overzealous cleanup may have broken something. I'm currently installing a SPARC system from scratch to try to prove out that I only get into this situation following the running of the AI server-side test suite. I expect that Beijing QE has successfully done installs using the 111b2 packages, but can't check with them at the moment due to the time zone difference. Has someone else here done x86 installs using 111b2? I want to make sure this problem is confined to machines where I've run the suite. Andre
