lee@doc.cs.unm.edu (Lee Ward) (02/22/89)
Any idea why a 3/60 client would panic trying to mount root from a 3/180
server?
It looks like the client responds with "bad port" with an ICMP error after
loading the kernel on the root mount request. The ethernet interface on
the 3/180 is configured:
ie0: flags=43<UP,BROADCAST,RUNNING>
inet 129.24.12.130 netmask fffff800 broadcast 129.24.15.255
All the 60's do this under any network load. Below is the (edited to
relevant info) output of an etherfind. Oh yeah, eventually (hours later)
the boot succeeds. So much for the network IS the computer...
The end of tftp boot and initial mount to get the kernel:
UDP from 129.24.12.160.1344 to 129.24.12.130.1060 12 bytes
UDP from 129.24.12.160.1344 to 129.24.12.130.1060 12 bytes
UDP from 129.24.12.160.1344 to 129.24.12.130.1060 12 bytes
60 arp 129.24.12.160 129.24.12.130
UDP from 129.24.12.160.1022 to 129.24.12.130.710 124 bytes
RPC Call prog 100026 proc 2 V1
60 arp 129.24.12.160 129.24.12.130
UDP from 129.24.12.160.1021 to 129.24.12.130.sunrpc 136 bytes
RPC Call portmapper PMAPPROC_GETPORT V2
UDP from 129.24.12.160.1020 to 129.24.12.130.697 128 bytes
RPC Call mountd MNTPROC_MNT V1
UDP from 129.24.12.160.1019 to 129.24.12.130.2049 124 bytes
RPC Call nfs RFS_GETATTR V2
After the kernel is loaded...
UDP from 129.24.12.160.1019 to 129.24.12.130.2049 136 bytes
RPC Call nfs RFS_READ V2
60 arp 129.24.12.160 129.24.12.130
UDP from 129.24.12.160.1023 to 129.24.12.130.710 124 bytes
RPC Call prog 100026 proc 2 V1
UDP from 129.24.12.160.1023 to 129.24.12.130.sunrpc 104 bytes
RPC Call portmapper PMAPPROC_GETPORT V2
UDP from 129.24.12.160.1023 to 129.24.12.130.697 128 bytes
RPC Call mountd MNTPROC_MNT V1
UDP from 129.24.12.160.1023 to 129.24.12.130.697 128 bytes
RPC Call mountd MNTPROC_MNT V1
UDP from 129.24.12.160.1023 to 129.24.12.130.697 128 bytes
RPC Call mountd MNTPROC_MNT V1
UDP from 129.24.12.160.1023 to 129.24.12.130.697 128 bytes
RPC Call mountd MNTPROC_MNT V1
UDP from 129.24.12.160.1023 to 129.24.12.130.697 128 bytes
RPC Call mountd MNTPROC_MNT V1
ICMP from 129.24.12.160 to 129.24.12.130 dst unreachable bad port
bad packet was: UDP from 129.24.12.130.697 to 129.24.12.160.1023 68 bytes
RPC Call prog 2 proc 10 V1129512960
ICMP from 129.24.12.160 to 129.24.12.130 dst unreachable bad port
bad packet was: UDP from 129.24.12.130.697 to 129.24.12.160.1023 68 bytes
--Lee (Ward)mcooper%acamar.usc.edu@oberon.usc.edu (Michael A. Cooper) (03/06/89)
We had a similiar problem. The boot program would get downloaded, the kernel would get load, autoconf, and then it would query the net again (net boot) asking for it's server. At that point several icmp destination unreachable messages would be sent by random hosts on the ethernet. This would cause the kernel to crash since it couldn't mount its root and swap. It turns out the "random hosts" had incorrectly configured their broadcast address and/or their subnet mask. Fixing these "random hosts" solved the problem. What's really weird is that when boot did a net boot query initially it also saw icmp dest unreachable messages, but it retried its query and eventually succeeded. The kernel appears to only try one query and then give up and die. Michael A. Cooper, University Computing Services, U of Southern California UUCP: ...!uunet!oberon!mcooper BITNET: mcooper@mouse ARPA: mcooper@oberon.USC.EDU PHONE: (213) 743-2957