Date: Thu, 25 Mar 2004 15:01:35 -0800 (PST) From: Steve Lin <stevenclin@yahoo.com> To: freebsd-questions@freebsd.org Subject: One link dropping on Multilink ppp Message-ID: <20040325230135.41941.qmail@web41012.mail.yahoo.com>
next in thread | raw e-mail | index | archive | help
Hello, I'm using user-land ppp and setting up a multi-link PPP over TCP connection, using "set device a.b.c.d:xxxx/tcp a.b.c.e:xxxx/tcp". When both a.b.c.d and a.b.c.e are reachable, everything is great - multilink works just fine. However, when one of my underlying links goes down after the multilink connection is established, the multilink ppp session appears to "hang" (no traffic gets through) for about 8-10 minutes, before it seems to realize that only one link is available and starts sending traffic through the one remaining link. I'm trying to reduce the 8-10 minute lag time before traffic continues flowing on the single remaining link, and I can't figure out what event or timer is occuring at this 8-10 minute mark to make things start working again. I've tried "enable lqr" and "set lqrperiod 10" and "set reconnect 10 9999999" and tweaking "set redial", to no avail. Here's the relevant config files for the client: -------------client ppp.conf---------------- default: ident user-ppp VERSION (built COMPILATIONDATE) set timeout 180 enable dns enable lqr accept lqr set lqrperiod 10 dual: set escape 0xff set device 63.220.17.114:6669/tcp 63.220.17.115:6669/tcp set phone set dial set login set authname testname set authkey testpass set enddisc magic set timeout 0 set ifaddr 10.0.4.2 10.0.4.1 nat enable yes set mrru 1500 clone 1,2 link 1 set device 63.220.17.114:6669/tcp link 2 set device 63.220.17.115:6669/tcp link deflink remove link 1,2 set reconnect 10 9999999 link 1,2 enable lqr link 1,2 accept lqr link 1,2 set lqrperiod 10 link 1,2 set redial 10 3 add default HISADDR link 1,2 set mode ddial -----------end client ppp.conf---------------- and server: ---------- server ppp.conf------------------- ppp-in: set timeout 0 set ifaddr 10.0.4.1 10.0.4.2 enable PAP set enddisc mac set mrru 1500 -----------end server ppp.conf---------------- Here's the log shown on the client from "set log local lcp phase ipcp": -----------client log------------------ PPP ON bondtest> set log local lcp ipcp phase >>> NOTE : Link 2 Dropped here <<< PPP ON bondtest> Phase: 2: ** Too many LQR packets lost ** LCP: 2: LayerDown LCP: 2: State change Opened --> Starting Phase: 2: open -> lcp LCP: 2: LayerFinish LCP: 2: State change Starting --> Initial Phase: 2: Disconnected! Phase: 2: lcp -> logout Phase: 2: Disconnected! Phase: 2: logout -> hangup Phase: 2: Connect time: 83 secs: 920 octets in, 5222 octets out Phase: 2: 16 packets in, 60 packets out Phase: total 74 bytes/sec, peak 567 bytes/sec on Thu Mar 25 17:47:47 2004 Phase: 2: hangup -> opening Phase: 2: Enter pause (10) for redialing. Phase: 2: Connecting to 63.220.17.115:6669/tcp Warning: 2: connect: Interrupted system call Phase: 2: Enter pause (10) for redialing. Phase: 2: Connecting to 63.220.17.115:6669/tcp Warning: 2: connect: Interrupted system call Warning: 2: Device (63.220.17.115:6669/tcp) must begin with a '/', a '!' or contain at least one ':' Phase: 2: Enter pause (10) for redialing. Phase: 2: Connecting to 63.220.17.115:6669/tcp Warning: 2: connect: Interrupted system call Warning: 2: Device (63.220.17.115:6669/tcp) must begin with a '/', a '!' or contain at least one ':' Phase: 2: Enter pause (10) for redialing. Phase: 2: Connecting to 63.220.17.115:6669/tcp Warning: 2: connect: Interrupted system call >>> This continues every 10 seconds indefinitely, even after the link is apparently re-established using only link 1 <<< ------------end client log--------------- I can provide the server log too but figured this is already long enough... Any ideas how I can get it to drop back to using just one link without waiting 10 minutes? Thanks, Steve Lin stevenclin@yahoo.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040325230135.41941.qmail>