*BUG REPORTED* Cannot configure VRRPD inside a VE [message #12853] |
Fri, 11 May 2007 17:42 |
oviney
Messages: 17 Registered: April 2007 Location: Ottawa
|
Junior Member |
|
|
Hi All:
I have recently been testing VRRPD () on a RHEL4 update 4 VE and ran into the following error when trying to configure...
Error:
<snippet>
cant open raw socket. errno=97. (try to run it as root)
</snippet>
Here is the command that I ran:
<snippet>
vrrpd -i eth0 -p 25 -v 1 172.17.8.181
</snippet>
So at a first guess, it would appear that the VE doesn't have promiscuous access to the Ethernet device.
Any help would be much appreciated.
[Updated on: Thu, 07 June 2007 07:43] by Moderator Report message to a moderator
|
|
|
|
|
|
Re: Cannot configure VRRPD inside a VE [message #13724 is a reply to message #12900] |
Mon, 04 June 2007 12:15 |
Tellerdreher
Messages: 13 Registered: October 2005
|
Junior Member |
|
|
I do not use vrrpd but keepalived and have the same problem. The process cannot open the socket to send the multicast VRRP keepalive messages:
Jun 4 14:10:21 keepalived1 Keepalived_vrrp: cant open raw socket. errno=97
[pid 32119] socket(PF_INET, SOCK_RAW, 0x70 /* IPPROTO_??? */) = -1 EAFNOSUPPORT (Address family not supported by protocol)
I am using 2.6.18-8.1.3.el5.028stab033.1
PS: /etc/protocols: 0x70 == 112 == VRRP
PPS: I think here is the "problem" (net/socket.c):
int vz_security_proto_check(int family, int type, int protocol)
...
switch (family) {
case PF_UNSPEC:
case PF_PACKET:
case PF_NETLINK:
case PF_UNIX:
break;
case PF_INET:
switch (protocol) {
case IPPROTO_IP:
case IPPROTO_ICMP:
case IPPROTO_TCP:
case IPPROTO_UDP:
case IPPROTO_RAW:
break;
default:
return -EAFNOSUPPORT;
}
break;
...
Is this really only a security check (you can only open sockets with protocl == 0(ip), 1(icmp), 6(tcp), 17(udp), 255(??) inside a VE)? Shouldn't this be configurable (with a Capability or so?)?
[Updated on: Tue, 05 June 2007 08:43] Report message to a moderator
|
|
|
|
|