Em 05-02-2014 07:09, Tomoyuki Sakurai escreveu: > hi misc, > > i'm trying to run OpenBSD with two default gateways, one for openvpn and > another for everything else. openvpn is in rdomain 1 and everything works > fine. > > OpenBSD 5.5-beta (GENERIC.MP) #284: Mon Feb 3 07:57:32 MST 2014 > t...@amd64.openbsd.org:/usr/src/sys/arch/amd64/compile/GENERIC.MP > > # cat hostname.em1 > !echo "starting em1" > rdomain 1 > group vpn > inet yy.yy.yy.yy/28 > !/sbin/route -T 1 add default yy.yy.yy.default > !echo -n "starting sshd in rdomain 1" > !route -T 1 exec /etc/rc.d/sshd start && echo "." > !echo -n "starting openvpn in rdomain 1" > !install -d -o _openvpn -g _openvpn -m 0755 /var/run/openvpn && > /sbin/route -T 1 exec /usr/local/sbin/openvpn --daemon --config > /etc/openvpn/server.conf && echo "." > > # cat hostname.tun0 > !echo "starting tun0" > up > 10.100.16.1 10.100.16.2 netmask 0xffffffff > !route add 10.100.16.0/24 10.100.16.2 > > however, openvpn cannot be restarted in multi-user mode. it hangs in > get_default_gateway(). > > http://www.openssh.com/cgi-bin/cvsweb/ports/net/openvpn/patches/patch-src_openvpn_route_c?rev=1.2;content-type=text%2Fplain > > ktrace shows that read(2) to routing socket does not return. > > 10068 openvpn CALL socket(PF_ROUTE,SOCK_RAW,0) > 10068 openvpn RET socket 3 > 10068 openvpn CALL sigprocmask(SIG_BLOCK,~0<>) > 10068 openvpn RET sigprocmask 0<> > 10068 openvpn CALL mprotect(0x10ee093000,0x2000,0x3<PROT_READ|PROT_WRITE>) > 10068 openvpn RET mprotect 0 > 10068 openvpn CALL mprotect(0x10ee093000,0x2000,0x1<PROT_READ>) > 10068 openvpn RET mprotect 0 > 10068 openvpn CALL sigprocmask(SIG_SETMASK,0<>) > 10068 openvpn RET sigprocmask ~0x10100<SIGKILL|SIGSTOP> > 10068 openvpn CALL write(0x3,0x10ee1949c0,0x80) > 10068 openvpn GIO fd 3 wrote 128 bytes > > "\M^@\0\^E\^D\0\0\0\0\0\0\0\0\^E\0\0\0\^C\0\0\0\0\0\0\0\0\0\0\0\^A\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\ > > \0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\^P\^B\0\0\0\0\0\0\0\0\0\0\0\0\0\0\^P\^B\ > \0\0\0\0\0\0\0\0\0\0\0\0\0\000" > 10068 openvpn RET write 128/0x80 > 10068 openvpn CALL sigprocmask(SIG_BLOCK,~0<>) > 10068 openvpn RET sigprocmask 0<> > 10068 openvpn CALL mprotect(0x10ee093000,0x2000,0x3<PROT_READ|PROT_WRITE>) > 10068 openvpn RET mprotect 0 > 10068 openvpn CALL mprotect(0x10ee093000,0x2000,0x1<PROT_READ>) > 10068 openvpn RET mprotect 0 > 10068 openvpn CALL sigprocmask(SIG_SETMASK,0<>) > 10068 openvpn RET sigprocmask ~0x10100<SIGKILL|SIGSTOP> > 10068 openvpn CALL read(0x3,0x10ee1949c0,0x260) > > when invoked from hostname.em1 during boot, read(2) immediately returned with > ESRCH. according to route(4), messages written to the socket should be > returned. how can read(2) be blocked? > the behavior is same on 5.4, 5.3 and current. > I've used rdomains, but not for this. In this case I would use mpath and pf only. I really do not see the need for using rdomains in this case. It introduces too much complexity for a simple thing.
Cheers, -- Giancarlo Razzolini GPG: 4096R/77B981BC
