From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.9) Gecko/20020412 Debian/0.9.9-6 Description of problem: Trying to start the netdump client on a machine that has eth2 configured but not eth0 fails. Version-Release number of selected component (if applicable): 0.6.6-1 How reproducible: Always Steps to Reproduce: 1. Install RHAS21 on a machine with several network cards 2. Configure eth2 only 3. "service netdump start" Actual Results: : cannot arp 172.22.48.89 /etc/init.d/netdump: Usage:: command not found /etc/init.d/netdump: propagate}: command not found /etc/init.d/netdump: line 4: 15602 Broken pipe initlog $INITLOG_ARGS -c "$*" /etc/init.d/netdump: line 4: 15605 Broken pipe initlog $INITLOG_ARGS -c "$*" : must be an ethernet interface Usage: service netdump {start|stop|status|restart|condrestart|propagate} initializing netdump /lib/modules/2.4.9-e.3smp/kernel/drivers/net/netconsole.o: invalid argument syntax for netdump_target_eth_byte0: 'x' /lib/modules/2.4.9-e.3smp/kernel/drivers/net/netconsole.o: insmod /lib/modules/2.4.9-e.3smp/kernel/drivers/net/netconsole.o failed /lib/modules/2.4.9-e.3smp/kernel/drivers/net/netconsole.o: insmod netconsole failed [FAILED] Expected Results: The netdump client should have started, prepared for dumping on eth2. Additional info: The network interface to dump on should be configurable in /etc/sysconfig/netdump.
Dave, can you look into this?
> Actual Results: : cannot arp 172.22.48.89 The error sequence is caused by the arping command in the netdump script: # fill the arp cache with needed data arping -c 1 $1 > /dev/null 2>&1 [ $? -ne 0 ] && echo "$prog: cannot arp $1" 1>&2 && usage What happens if you enter "arping -c 1 172.22.48.89" on a command line?
[root@borg root]# arping -c 1 172.22.48.89 Interface "eth0" is down [root@borg root]# ifconfig eth2 Link encap:Ethernet HWaddr 00:06:5B:3D:E5:58 inet addr:172.22.48.108 Bcast:172.22.48.255 Mask:255.255.255.0 UP BROADCAST NOTRAILERS RUNNING MTU:1500 Metric:1 RX packets:6053191 errors:0 dropped:0 overruns:0 frame:0 TX packets:6103716 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100 RX bytes:2058350921 (1962.9 Mb) TX bytes:2149915367 (2050.3 Mb) Interrupt:16 Base address:0x4000 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:1172727 errors:0 dropped:0 overruns:0 frame:0 TX packets:1172727 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:279730513 (266.7 Mb) TX bytes:279730513 (266.7 Mb) [root@borg root]# arping -I eth2 -c 1 172.22.48.89 ARPING 172.22.48.89 from 172.22.48.108 eth2 Unicast reply from 172.22.48.89 [00:02:55:AC:05:74] 0.726ms Sent 1 probes (1 broadcast(s)) Received 1 response(s) [root@borg root]# rpm -qf `which arping` iputils-20001110-6 Thus, a simple workaround (that I've applied) is to hard code -I eth2 into /etc/init.d/netdump. It works, but is of course not a long term solution.
> The network interface to dump on should be configurable in > /etc/sysconfig/netdump. Thanks -- that's the intent of the DEV= line in /etc/sysconfig/netdump, but unfortunately it's ignored by the arping command. Just for sanity's sake, could you verify that this works: 1. Set "DEV=eth2" in /etc/sysconfig/netdump 2. Set "DEV=eth0" in /etc/rc.d/init.d/netdump (as the default that will get overwritten by "eth2") 3. Change your workaround line to: arping -c 1 -I $DEV $1 > /dev/null 2>&1 Thanks very much.
Yup, that seems to work.
Reminder: this patch has been checked into base; needs to be part of AS errata. --- netdump/netdump.init.orig Mon Jul 15 11:24:35 2002 +++ netdump/netdump.init Mon Jul 15 11:24:55 2002 @@ -28,7 +28,7 @@ # Default values LOCALPORT=6666 -DEV= +DEV=eth0 NETDUMPADDR= NETDUMPMACADDR= NETDUMPPORT=6666 @@ -66,7 +66,7 @@ local host=$1 local oldIFS arp_output line - arping -c 1 $host &> /dev/null + arping -c 1 -I $DEV $host &> /dev/null [ $? -ne 0 ] && echo "$prog: cannot arp $host" 1>&2 && usage # output from arp -a of the form:
*** Bug 73629 has been marked as a duplicate of this bug. ***
Another workaround is simply to comment out the line: [ $? -ne 0 ] && echo "$prog: cannot arp $host" 1>&2 && usage
This will be addressed in the 0.6.11-1 release in Update 3.