1

I have a server installed Debian Xen 4.1. I will refer to this server later with letter A. The server has two virtual servers running. Let's call these servers with letters B and C.

All servers A(host), B(vm), and C(vm) has public ips (they have only public ips, nothing else) Server A: x.x.x.82 Server B: x.x.x.83 Server C: x.x.x.84

They all have the same netmask and the same gateway. Server A has interfaces xenbr0, vif2.0 and vif3.0. The server B has an interface eth0 (linked to vif2.0). The server C has an interface eth0(linked to vif3.0). I've checked routes and gateways and netmasks are correct. And the configuration works.

route -n (on server A)
Destination     Gateway         Genmask         Flags Metric Ref    Use 
Iface
0.0.0.0         x.x.x.81        0.0.0.0         UG    0      0        0 xenbr0
x.x.x.80        0.0.0.0         255.255.255.240 U     0      0        0 xenbr0

route -n (on server B and C)
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         x.x.x.81        0.0.0.0         UG    0      0        0 eth0
x.x.x.80        0.0.0.0         255.255.255.240 U     0      0        0 eth0

This may work days, weeks, months but then something happens. Just randomly the virtual server B cannot reach gateway anymore, and no-one cannot access to the server B via public IP. All interfaces are up, routes are still corrects. Server A and C can ping B, and B can ping them. The gateway is unreachable. I've checked iptables on the host (server A), and it's absolutely correct and nothing hasn't changed. I've read dmesg, syslog and all the log files I could ever imagine on the host server A and the virtual machine B - but cannot find any releated error. Routing just doesn't work anymore as it should! And nothing has been changed either!

As far as I know, only solution is to reboot virtual machine. After that it works perfectly next few months.

I've read maybe hundreds forums and discussions but haven't find the solution - usually they have something wrong with the configuration or they get some error message. When the server B cannot connect to the gateway anymore, the server C works perfectly fine and their network settings are identical (except different ip of course).

I've checked also brctl, and it states that STP enabled is "no". Can this has anything to do with that? It's hard to believe because VM C is still working perfectly.

I would appreciate any suggestions where to start to look the reason for this. I've checked iptables, ifconfig, brctl, dmesg, logs etc. so many times that I'm out of ideas.

0

You must log in to answer this question.

Browse other questions tagged .