Ethernet interface problems on Solaris 8.
I'm having some problems with a Solaris 8 on x86 server not seeing its ethernet interface between reboots without reconfiguring every time.
About two weeks ago, a backup script on my main file sever spontaneously appeared to crash NIS, and the box was rebooted. When it came back up, it appeared to have forgotten about its ethernet interface, an Intel GigE card at e1000g0. The error message at boot was:
"Protocol wrong type for socket"
... followed by the attempt to move the IP from e1000g0 to any other active interface, which of course failed, since it's the only interface in the box.
I've found that the only way to get the interface up is to boot -rs so it forces a reconfiguration and then drops into single-user mode. Then I can plumb the interface and number it by hand. I can then init 3 and everything is fine. If I don't boot into single user mode, it tries to start NIS, fails, and the box hangs.
If I reboot at this point, the box fails detect the interface again and I have to go through the whole process once more.
I've tried at least one other Intel GigE card, as well as moving them between PCI slots to no avail. I've removed the driver and added it back in, same thing.
Does anyone have any idea what might be the problem? I'm completely stumped.
I'm not sure if it's pertinent, but the when I run my backup script and it crashes NIS, I get the following errors in /var/adm messages.
Aug 10 15:27:10 myserver smbd[23184]: [ID 105279 user.error] nis_cast: t_open: /dev/udp:No such device or address
Aug 10 15:27:10 myserver smbd[23176]: [ID 566193 user.error] n2a get_local_info: ioctl (get interface flags): No such device or address
This is what happened to cause the reboot in the first place, after which the interface failed to show up. I'm not sure if it's important, but I figured I'd include it anyway.
Thanks for any help!
Franklin
|