[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Reboots early 9/26/99
- To: outages-list@eskimo.com
- Subject: Reboots early 9/26/99
- From: Robert Dinse <nanook@eskimo.com>
- Date: Sun, 26 Sep 1999 15:45:37 -0700 (PDT)
- Resent-Date: Sun, 26 Sep 1999 15:45:39 -0700
- Resent-From: outages-list@eskimo.com
- Resent-Message-ID: <"rV_cT.0.Rf2.J8gxt"@mx1>
- Resent-Sender: outages-list-request@eskimo.com
I rebooted tia1 a number of times early this morning in an attempt to
troubleshoot a strange NIS problem which unfortunately remains unresolved.
Technical explanation follows:
On this box, tia1 fails to bind with a ypserver via broadcast, it finds a
server with ypset (and this is normally the way it functions anyway), but the
fact that the former doesn't work and it does work on the other machines is
troublesome.
I went over the initialization scripts, they are same as on other
machines, copied the binaries for ypserv and ypbind thinking maybe they had
been mangled, (and they were old versions with some known security problems),
but to no avail.
The problem manifests itself in that the ypset command will not work with
a specified hostname because the bind using broadcast for a ypserver failed, so
it doesn't resolve the hostname 'tia1' (itself).
Now, there are really two mysteries here; why the bind using broadcast
fails, and why it can't resolve the host name when tia1 is in /etc/hosts AND
the /etc/host.conf says to use hosts before NIS. The ypbind binary there is
statically linked, so even if the dynamic resolver libs were munched this
should still worked.
It is one of three Sun4C machines, and the only one of the three running
NIS. These machines are generally unstable but this particular machine has
been stable. They have the Wytek Power-uP chips in them which apparently had
some bugs, but none of that explains this behavior and I have to say at this
point I am completely stumped.