stefan@lbl-csam.arpa (Stefan gottschalk) (10/28/89)
After rebooting the NeXT, the system forget about all the accounts, hosts, file systems, and other whatnot contained in the NetInfo database. Fortunately, the system resorted to the flat ascii files so I was able to rig myself up an account and begin looking for the cause of the problem. However, after a couple of days of reading about NetInfo, the booting procedures, and the various daemons and services which must be in place, I still haven't any idea as to what went wrong. Here are some clues: At the end of the execution of rc.boot, three error messages appear on the screen in rapid succession. An instant later the screen blacks out and the login window appears, but I believe they are as follows: npd[108]: NetInfo Problem: Communication failure npd[108]: attempting to auto register printer npd[108]: NetInfo Problem: Communication failure These messages also appear with various prefixes in the file private/adm/messages (or /usr/adm/messages, take your pick). There is a 'core' file in the /etc/netinfo directory. I'm told that dbx can look at coredumps intelligently (if the failing program was compiled with the '-g' option), but I'm not familiar with dbx or debugging with core dumps, so I haven't tried that yet. 'ps -x' shows nibindd, the guy who creates the netinfod's, but the netinfo daemons themselves aren't there. 'niutil -list . /' produces the message "niutil: can't open .:/" The man page for nibindd says that it automatically brings up a netinfo daemon for every *.nidb directory in /etc/netinfo. So I killed the old nibindd (with 'kill -QUIT' which seemed to work), and started up a new one. Sure enough, 'ps -x' showed a 'usr/etc/netinfod local' whose parent process was nibindd, which was a child process of the actual nibindd that I started. But niutil still came back with the "can't open" message. I figured maybe the local.nidb directory was messed up, so I tried copying the local.nidb directory from the 'Software Release 1.0' optical over the the working version on the troubled NeXT. That didn't help. No change in behavior. That really worries me. I figure that if it's not the database, then it is the environment causing trouble, and if that's the case then I imagine the problem could be anywhere! Appendix A of the System Administration Guide, "System Initialization," p. A-18 states that NetInfo depends upon the portmap daemon. Good ol' 'ps -x' shows that portmap is indeed present. I checked the permissions on the '.nidb' files to make sure they were identical to those on the optical. They were. Furthermore, everything I checked _is_ in the correct place in the file system. The contents of '/etc/hostconfig', which the rc and rc.local scripts source, are correct, compared to the version on the optical. Note: I am certain that the optical disk has not been altered. The first thing we did on receipt of 1.0 was rebuild the SCSI from the optical, which we then put into storage. So there it is. I may wind up simply rebuilding the disk after saving the important files, but I'd like to know what happened so I can avoid making the same mistake again. Probably I zapped some file somewhere while romping around as root, but I'll be damned if I know which one! Please note that this is NOT a complaint (it's a plea for help). The NeXT is a little frustrating, at times, because of its differences from BSD 4.3. But it is these differences that make the NeXT such an interesting and exciting machine! Well, if anyone has any suggestions (any at all!) then please don't hesitate to send me a note. I'd appreciate it. Thanks in advance, Stefan Gottschalk (stefan@csam.lbl.gov)
phd_ivo@gsbacd.uchicago.edu (10/28/89)
In article <4038@helios.ee.lbl.gov>, stefan@lbl-csam.arpa (Stefan gottschalk) writes... > <variety of netinfo problems, basically dead netinfo> I had the same problems once after fooling around with both netinfo and flat files alternately. I ultimately resorted to rebuilding my hard-disk. At the time, I posted a note to comp.sys.next which said that netinfo should have a way of just saving everything, so that an administrator can play with netinfo for a while, and if he/she screws up, simply restore the old settings and reboot. I am now very careful about NetInfo, as I fear the symptoms could return. Thus, if anyone knew the answer to how to fix this problem, I would also appreciate the answer. /ivo