Re: 3.8 bridge trouble
* Pailloncy Jean-Gerard [EMAIL PROTECTED] [2006-02-15 21:27]: I add asked Henning at EuroBSDCon'2005 about this, and he says to me that if I could avoid this setup (bridge) and use a router: this is the way to go. actually, I am pretty certain I told you you'll need stp at least. I did and still do recommend routing and not bridging in any case. -- BS Web Services, http://www.bsws.de/ OpenBSD-based Webhosting, Mail Services, Managed Servers, ... Unix is very simple, but it takes a genius to understand the simplicity. (Dennis Ritchie)
Re: 3.8 bridge trouble
* Pailloncy Jean-Gerard [EMAIL PROTECTED] [2006-02-16 10:26]: Nest tr y: I setup the two nics to be in 10bt mode and not in 100bt. The box freezes, all the segment go down. Near nothing comes in or out, from any other serveres of the segment. answer to ssh was with a lag of few minutes (for control-C) I just unplug the box, and stop putting down my network. sounds like you;re way overloading the tiny CPUs -- BS Web Services, http://www.bsws.de/ OpenBSD-based Webhosting, Mail Services, Managed Servers, ... Unix is very simple, but it takes a genius to understand the simplicity. (Dennis Ritchie)
Re: 3.8 bridge trouble
On Wed, 15 Feb 2006, Pailloncy Jean-Gerard wrote: Second part of the test, I set up a bridgename.bridge0 file with the 2 nics up with STP, and I restart the soekris. Few seconds after the end of the boot (login prompt) immediate reboot of the soekris. I stop it by, as soon as login prompt appears, to log in and put down the bridge. In fact sometimes when there is a big storm the soekris reboots too. Is the watchdog timer (sysctl kern.watchdog) set? I've seen Soekrises reboot because of that when under high network load. The reboot happens with kern.watchdog.auto=0 and with kern.watchdog.auto=1 !!! Nest tr y: I setup the two nics to be in 10bt mode and not in 100bt. The box freezes, all the segment go down. Near nothing comes in or out, from any other serveres of the segment. answer to ssh was with a lag of few minutes (for control-C) I just unplug the box, and stop putting down my network. Cordialement, Jean-Girard Pailloncy
Re: 3.8 bridge trouble
So something like this? [ISP] | [SW1][SW2] || | +--+ [SKR] Yes. I want something redundant, there is a loop somewhere. I want to test the soekris as a bridge with 3 links: one to my provider, one to each switch. I start the test by setting a bridge on 2 nics of the soekris connected to the 2 switchs. So now there is a loop between the switchs and the soekris. And the first problem appears. If I ping from any computers on the segment to anything else I get a storm of thousand packet/ seconds until I unplug a cable from the loop. That's because of the loop. Even if the switches cache ARP replies and send further packets directly to their destinations, the initial broadcasted ARP request is going to go round and round. So I test with/without STP, link0: no more luke. I test with STP first, then without: same problem. I try with link0 Setting this flag stops all IP multicast packets from being forwarded by the bridge. To prevent the initial broadcast to loop. Does not work... It a broadcast and not a multicast, after all. Maybe I need a rule to filter all packet from the oekris itself. Cordialement, Jean-Girard Pailloncy
3.8 bridge trouble
Hi, I want to use 2 soekris as redundant bridge filter, but I have many troubles. I installed on soekris with 128 Mo SANDISK and a stock version of 3.8. I modify the setup to have a readonly root, 2 mfs partitions (dev, var), syslog to buffer and to remote loghost. Everything runs nice. Every servers I have have at least 2 nics. I connect each server to 2 switchs, and the switchs together. And for the moment only one switch has an uplink to my provider. I will have (soon) a second link to the second switch. I want to test the soekris as a bridge with 3 links: one to my provider, one to each switch. I start the test by setting a bridge on 2 nics of the soekris connected to the 2 switchs. So now there is a loop between the switchs and the soekris. And the first problem appears. If I ping from any computers on the segment to anything else I get a storm of thousand packet/seconds until I unplug a cable from the loop. So I test with/without STP, link0: no more luke. Second part of the test, I set up a bridgename.bridge0 file with the 2 nics up with STP, and I restart the soekris. Few seconds after the end of the boot (login prompt) immediate reboot of the soekris. I stop it by, as soon as login prompt appears, to log in and put down the bridge. In fact, each time all the cables is connected to have a loop, in the 15 seconds I put up the bridge the box reboots (may be the same delay as to start forwarding packet). But if I put up the bridge with only one cable and put the second ones after not much problems. In fact sometimes when there is a big storm the soekris reboots too. I add asked Henning at EuroBSDCon'2005 about this, and he says to me that if I could avoid this setup (bridge) and use a router: this is the way to go. OK. But My provider gives me a block of 256 IPs with two gateways and before asking him for something else I would like to know what I can do. May be the new coming version 3.9 have some new stuff. I do not know. Any help to have a running setup is welcomed, and to debug the unexpected reboot. Cordialement, Jean-Girard Pailloncy
Re: 3.8 bridge trouble
Pailloncy Jean-Gerard wrote: I want to use 2 soekris as redundant bridge filter, but I have many troubles. [...] Every servers I have have at least 2 nics. I connect each server to 2 switchs, and the switchs together. And for the moment only one switch has an uplink to my provider. I will have (soon) a second link to the second switch. So something like this? [ISP] | [SW1][SW2] || | +--+ [SKR] I want to test the soekris as a bridge with 3 links: one to my provider, one to each switch. I start the test by setting a bridge on 2 nics of the soekris connected to the 2 switchs. So now there is a loop between the switchs and the soekris. And the first problem appears. If I ping from any computers on the segment to anything else I get a storm of thousand packet/seconds until I unplug a cable from the loop. That's because of the loop. Even if the switches cache ARP replies and send further packets directly to their destinations, the initial broadcasted ARP request is going to go round and round. So I test with/without STP, link0: no more luke. The above test was without STP, yes? Are you saying that when you enabled STP you got a message of link0: no more luke? Do your switches support STP and have it enabled? Second part of the test, I set up a bridgename.bridge0 file with the 2 nics up with STP, and I restart the soekris. Few seconds after the end of the boot (login prompt) immediate reboot of the soekris. No idea what's happening here. Best guess would be protection against overheating. If there's nothing in your OS logs, check your BIOS for any management logs. -- Jason Stubbs
Re: 3.8 bridge trouble
On Wed, 15 Feb 2006, Pailloncy Jean-Gerard wrote: Second part of the test, I set up a bridgename.bridge0 file with the 2 nics up with STP, and I restart the soekris. Few seconds after the end of the boot (login prompt) immediate reboot of the soekris. I stop it by, as soon as login prompt appears, to log in and put down the bridge. In fact sometimes when there is a big storm the soekris reboots too. Is the watchdog timer (sysctl kern.watchdog) set? I've seen Soekrises reboot because of that when under high network load. -- Cam