Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 07 Aug 1997 10:56:49 -0700
From:      John Milford <jwm@CSUA.Berkeley.EDU>
To:        hackers@freebsd.org
Subject:   Routing problems
Message-ID:  <199708071757.KAA24249@soda.CSUA.Berkeley.EDU>

next in thread | raw e-mail | index | archive | help

	A new 2.2.2 machine has recently been droping off the net
for some unknown reason.  This can take from 15 min. to 24 hour to
happen and the only way to bring it back is to reboot.   At least that
is the only way I have found.  I have found that the routing table
looks a little strange when this happens:

# netstat -rn
Routing tables

Internet:
Destination        Gateway            Flags     Refs     Use     Netif
Expire
default            128.32.240.254     UGSc        8       28      fxp0
127.0.0.1          127.0.0.1          UH          0     5640       lo0
128.32.134/24      link#1             UC          0        0
128.32.134.5       8:0:5a:47:21:53    UHLW        1      209       vx0   465
128.32.134.17      0:20:af:d3:78:9a   UHLW        1       26       lo0
128.32.134.240     0:0:f8:1:8e:e2     UHLW        0        7       vx0   779
128.32.134.254     0:0:93:10:6a:d8    UHLW        0        0       vx0  1116
128.32.240/24      link#2             UC          0        0
128.32.240.39      link#2             UHLW        0        3
128.32.240.190     0:a0:c9:6b:12:b1   UHLW        0    17276       lo0
128.32.240.231     link#2             UHLW        1      267
128.32.240.254     link#2             UHLW        8        0


	Where "link #2"  appears I would expect to see a MAC adress.
I would also expect to see all 240 net routes associated with fpx0.

	 I tried flushing the routing table:

# route flush
default              inr-115-eecs-new-net done
route: write to routing socket: No such process
got only -1 for rlen

# route -v flush
Examining routing table from sysctl
RTM_GET: Report Metrics: len 132, pid: 0, seq 0, errno 0,
flags:<UP,GATEWAY,STAT
IC,PRCLONING>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY,NETMASK>
 default inr-115-eecs-new-net-1.Berkeley.EDU (255) ffff 0 100
RTM_DELETE: Delete Route: len 132, pid: 0, seq 0, errno 0,
flags:<UP,GATEWAY,STA
TIC,PRCLONING>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY,NETMASK>
 default inr-115-eecs-new-net-1.Beecs-new-net-1.Berkeley.EDU (255)
ffff 0 100
RTM_GET: Report Metrics: len 124, pid: 0, seq 0, errno 0,
flags:<UP,HOST,LOCAL>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY>
 localhost localhost
RTM_GET: Report Metrics: len 136, pid: 0, seq 0, errno 0,
flags:<UP,CLONING>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY,NETMASK>
 128.32.134.0  (255) ffff ffff ff
RTM_GET: Report Metrics: len 128, pid: 0, seq 0, errno 0,
flags:<UP,HOST,LLINFO,
WASCLONED>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY>
 godzilla 8.0.5a.47.21.53
RTM_GET: Report Metrics: len 128, pid: 0, seq 0, errno 0,
flags:<UP,HOST,LLINFO,
WASCLONED,LOCAL>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY>
 fantasia 0.20.af.d3.78.9a
RTM_GET: Report Metrics: len 128, pid: 0, seq 0, errno 0,
flags:<UP,HOST,LLINFO,
WASCLONED>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY>
 saidin 0.0.f8.1.8e.e2
RTM_GET: Report Metrics: len  Report Metrics: len 128, pid: 0, seq 0,
errno 0, f
lags:<UP,HOST,LLINFO,WASCLONED>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY>
 inr-116-eecs.Berkeley.EDU 0.0.93.10.6a.d8
RTM_GET: Report Metrics: len 124, pid: 0, seq 0, errno 0,
flags:<UP,GATEWAY,HOST
,WASCLONED,PROTO3>
locks:  inits: <pksent,sendpipe,recvpipe,expire>
sockaddrs: <DST,GATEWAY>
 garnet.Berkeley.EDU inr-115-eecs-new-net-1.Berkeley.EDU
route: write to routing socket: No such process
got only -1 for rlen

	I have been looking at the source, but have not come up with
anything.  I have found that with just the 3c590 (vx0) that this does
not occur, and in fact that when this problem occurs the machine can
be acessed from the 134 net (vx0).

	Is it possible that the Intel card is causing the routing
table to get munged like this?  Any ideas on how to debug this will
be graetly appreciated.


		--John



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199708071757.KAA24249>