SIOCADDRT Invalid argument during boot

SIOCADDRT Invalid argument during boot

Post by Chris Murph » Tue, 19 Jan 1999 04:00:00



I have had this problem (<SIOCADDRT Invalid argument during boot>), and
seen various references to it in newsgroups. Just want to let anyone
who's interested know how I seem to have got rid of it. I am running
Redhat 5.2 on two machines linked via tcpip/ethernet. These machines are

named client and server. After doing a
RedHat-yes-I-want-networking-install, with the default gateway set
blank, getting them to talk with telnet and ftp was relatively easy. The

only problem was <SIOCADDRT Invalid argument during boot>.
    From my way of thinking (I'm not a networking person) in this setup
you do not want a default gateway, as these machines do not have packets

floating thru them that are destined for some unknown address. This was
bourne out to be the right way of thinking when I did another install,
letting the RedHat install default in a default gateway IP address - I
couldn't get client to talk to the outside world via PPP - I got the
error message "not replacing existing default route to eth0". However
the SIOCADDRT problem had disappeared. Thus the SIOCADDRT argument that
is wrong 'must' be the default gateway!
    There is a file called /etc/sysconfig/network which the install
seems to populate the last two lines of (one line refers to the gateway
address, the other the ethernet card). Now that I was 'sure' I didn't
need a default gateway it occured to me that the SIOCADDRT problem might

be due to the RedHat install putting a default gateway address of
0.0.0.0 into /etc/sysconfig/network when you blank out the field during
the install - SIOCADDRT (whatever it does!) might be croaking on a
0.0.0.0 arguement. To test this I commented out the last two lines of
/etc/sysconfig/network and rebooted. The SIOCADDRT boot problem did not
come back. A few other networking problems I was having also seemed to
disappear:
    -> When trying to connect via PPP, with the nameserver addresses
set, I would get this error (/var/log/messages):
        <
        CCP: timeout sending Config-Requests
        Connection terminated
        Receive serial link is not 8-bit clean:
        Problem: all had bit 7 set to 0
        Exit.
        >
        This error meant that you couldn't browse sites using their
names, only their addresses.
    -> When using ftp, this was sent to stderr: <ftp: bind: Address
already in use>
    If I could have gone about trying to solve this problem in a
slightly more
efficient way then I would certainly appreciate the advice!