Koozali.org: home of the SME Server
Obsolete Releases => SME 7.x Contribs => Topic started by: soprom on December 16, 2008, 09:15:39 PM
-
A server-and-gateway installation frequently has dhcp problem breaking lan access to workstations.
This is with smeserver-openvpn-bridge-fws-1.1-2.noarch.rpm
The symptoms are:
* dhcpd
Not configured to listen on any interfaces!
* pptp
GRE: read(fd=7,buffer=8056600,len=8260) from network failed: status = -1 error = Protocol not available
With openvpn, the InternalInterface Name must be br0
db configuration getprop InternalInterface Name
Faced with these problems I did:
yum remove pptpd e-smith-pptpd
yum install pptpd e-smith-pptpd
and discovered that InternalInterface was set back to eth0 instead of br0.
Does someone experienced this behaviour?
-
Following up...
pptp problem might also be linked to affa.
When a client tries to connect during an affa scheduled job, this error occurs:
Dec 16 16:00:08 serveur /sbin/e-smith/affa[25061]:.../users2affa: ...|valid|yes
Dec 16 16:01:35 serveur pptpd[25149]: CTRL: Client 192.197.166.31 control connection started
Dec 16 16:01:35 serveur pptpd[25149]: CTRL: Starting call (launching pppd, opening GRE)
Dec 16 16:01:35 serveur pppd[25150]: Plugin radius.so loaded.
Dec 16 16:01:35 serveur pppd[25150]: RADIUS plugin initialized.
Dec 16 16:01:35 serveur pppd[25150]: pppd 2.4.4 started by root, uid 0
Dec 16 16:01:35 serveur kernel: divert: not allocating divert_blk for non-ethernet device ppp0
Dec 16 16:01:35 serveur pppd[25150]: Using interface ppp0
Dec 16 16:01:35 serveur pppd[25150]: Connect: ppp0 <--> /dev/pts/2
Dec 16 16:01:35 serveur pptpd[25149]: GRE: read(fd=7,buffer=8056600,len=8260) from network failed: status = -1 error = Protocol not available
Dec 16 16:01:35 serveur pptpd[25149]: CTRL: GRE read or PTY write failed (gre,pty)=(7,6)
Dec 16 16:01:35 serveur pptpd[25149]: CTRL: Reaping child PPP[25150]
Dec 16 16:01:35 serveur pppd[25150]: Modem hangup
Dec 16 16:01:35 serveur pppd[25150]: Connection terminated.
Dec 16 16:01:35 serveur kernel: divert: no divert_blk to free, ppp0 not ethernet
Dec 16 16:01:35 serveur pppd[25150]: Exit.
-
Hi. It took me a while to track this problem. It occures in general each time you run a yum command. In fact, the script /etc/e-smith/events/actions/initialize-default-database will reset the name of the InternalInterface to eth0 (or eth1), and this script is allways called after yum.
I'm finishing a total rewrite of this contrib, I'll soone annonce it. Waiting for it, you can correct this problem with this:
vim /etc/e-smith/db/configuration/migrate/26openvpn-fix-internal-interface
And put this code :
{
use esmith::util;
my $vpn = $DB->get('openvpn-bridge');
my $status = $vpn->prop('status') || 'disabled';
my $bridge_name = $vpn->prop('brIf');
my $internal = $DB->get('InternalInterface');
my $internal_name = $internal->prop('Name');
if (($bridge_name eq $internal_name) || ($status eq 'disabled')){
return
}
$internal->merge_props('Name',$bridge_name);
}
This will fix the problem waiting for the new version.
Cheers
-
To add extra info:
I have this contrib installed on 6 servers and problems occur only on the one running the dhcp server.
-
GRE: read(fd=7,buffer=8056600,len=8260) from network failed: status = -1 error = Protocol not available
This was solved by recreating a new connection on the client side.