Thanks & monitoring interfaces

[mhd]

After our (failover-less) Cisco firewall and main router was behaving increasingly flakey last Friday, it was up to poor old programmer me to come up with a quick replament. I commandeered an old HP server (the only thing with enough ports) and tried to get some kind of firewall/router distribution going. Easier said than done, as installing the base package seemed hard to impossible for most rivals (something with the RAID controller, I guess). So, thank you, ebox developers for that. I had to burn some midnight oil, but I finally was able to get a basic system going.

But now I'm running into some problems with one of the interfaces. I noticed several times that it basically lost all connections to other interfaces, yet within the subnet (i.e. switch) itself, everything was working ok. To me this would point either towards the gateway or the switch itself. I thought about something hogging the bandwidth, but it's either rather fast, or no connection at all...

I haven't seen anything relevant in the log files, so I wondered whether there's something I could turn on the main ebox interface, to have a more detailed interface/connection log. The dashboard graphs aren't really enough for a detailed analysis.

So I'd be happy for any suggestions to help me get rid of this ugly glitch.

[azureice]

I have something similar happening.  One of the interfaces basically "dies". Everything within the subnet works but the connection between them is dead.  For instance, I have eth0 set to my internal network with an IP of 10.0.0.1.  For seemingly no reason, the connection is lost.  And oddly, the dashboard then reports that the IP of eth0 is 10.0.0.253, as if it acquired a DHCP address.  I have to manually set the IP to something else, then back to 10.0.0.1 and it will work again for another few minutes before it reverts back.

[Svein Wisnaes]

Mhd,

We are very happy to hear about your success so far.

To be able to help you efficient, we would like to know a bit more about your setup. Please check the link in my signature to get some pointers. The first we need is the version of eBox you are using.

[mhd]

Ebox 1.4-2, fresh download this weekend. Update and Upgrade were done, no reboot this far, though.

For ad-hoc monitoring, I've used jnettop and ntop in the meantime, although an ebox module providing similar data would be nice to have (rrdtool-based?).

I've been able to reproduce the problem since a few minutes, actually. It seems whenever a client acquires a new DHCP address, the interface loses its connectivity (ISC DHCPD 3.0.6, ebox-dhcp 1.4.2). Very weird. I've increased the default and maximum lease time to a few days, hoping to minimize that. Don't know whether the GUI can handle different dhcp daemons (or even major version upgrades of the ISC one).