On Thu, 05 Jun 2008 10:03:09 -0700, Andrew Sackville-West wrote: > On Thu, Jun 05, 2008 at 11:11:55AM -0400, [EMAIL PROTECTED] wrote: >> I have my network front end running Debian sarge (yet, it's time to >> upgrade at lest to etch). It's connected to the rest of the net by a >> DSL line. I've set up port-forwarding to selected machines on my LAN >> for the convenience of certain games, and bittorrent, and I'd like to >> use it for some always-on applications. >> >> But every now and then, once every week or two, the port-forwarding >> drops out, and I have to reestablish it by running the same script that >> gets run on startup. >> >> Does anyone have any idea how to track down the problem? I've looked at >> the system logs and not found anything suspicious. But probably I don't >> know which logs to look in, or what to look for in them, and it would be >> as plain as day if I did. >> >> If this is a known sarge problem, of course upgrading to etch would be >> the solution. > > just a couple of suggestions. > > 1) if you're planning to upgrade, go ahead and do it, the problem may > just go away. That doesn't really necessarily solve the problem > though... > > 2) does your DSL connection go down from time to time? maybe it's > dropping out and the system isn't recovering properly. > > 3) next time it goes down, instead of just rerunning the network > scripts, see if you can diagnose the situation a little bit. look at > the output of > > ifconfig > route > and maybe get a dump from iptables. > > then compare to the state of things when the system is working. > > A
OK. It happened again, just now. My son complained that the internet was down. I ssh'ed to my gateway machine, and ppp0 was up, but had a suspiciously low packet count (about 150 packets) within a few seconds it was up to about 250, so it looked as if ppp0 had just come up. I checked my iptables, and the port forwarding had shut down. I ran the script is uses to start port-forwarding at boot time, and it was up and running again in no time. I went and looked at syslog. It looks very much as if the ppp0 connection had shut down for a minute or two, and was automatically restarted. Apparently shutdown of ppp0 was enough to kill all the port-forwarding instructions relating to it (which was all of them), and the script did not get run when the ppp0 connection was reestablished. So, the question becomes, (a) how do I ensure that the script gets rerun when ppp0 comes up or (b) How do I keep the forwarding from being dropped when ppp0 goes down? -- hendrik -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
