[Lxc-users] Container inside an ESX VM

[Mauras Olivier]

Hello,

I'm struggling for two days now with some completely weird network
behaviours.
My host is a virtual machine hosted on an ESX farm. I planned to deploy
several containers on it to achieve various tasks.

Host is running Scientific Linux 6 with default kernel (2.6.32), and my
container is an Oracle Linux 6. I discovered that i had to change ESX
vswitch settings to allow promiscuous mode in order to make the host bridge
correctly behave, but it still gives me weird results.
Most of the time after having started the container, network inside the
container is erratic. I can ping or ssh from the host to the container, but
nothing gets out of the container or in the container from the LAN. While
the container is still running, if i issue a network restart on the host,
the container start behaving correctly and network works again as expected.
The problem is that it's not reliable at all. If i stop/restart the
container several times, it starts losing network again that i can only get
back by issuing the network restart on the host...

Here's my container configuration:
lxc.utsname = ct-011
lxc.network.type = veth
lxc.network.flags = up
lxc.network.link = br0
lxc.network.name = eth0
lxc.network.mtu = 1500
lxc.network.ipv4 = 0.0.0.0
lxc.mount = /etc/lxc/ct-01.fstab
lxc.rootfs = /srv/lxc/ct-01/

lxc.cap.drop = sys_module mknod
lxc.cap.drop = mac_override sys_time
lxc.cap.drop = setfcap setpcap sys_boot

I set the network from inside the container to avoid having to modify too
much of container init - I also tried setting IP from lxc config and it gave
me the same result.

My bridge is set with forward delay to 0 and STP on as having it disabled
doesn't work at all.

I don't have that much errors that could lead me to a solution here's a
snippet of my dmesg after restarting twice the network on the host:
e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
br0: starting userspace STP failed, starting kernel STP
br0: topology change detected, propagating
br0: port 1(eth0) entering forwarding state
device vethAuDQzn entered promiscuous mode
br0: topology change detected, propagating
br0: port 2(vethAuDQzn) entering forwarding state
br0: port 2(vethAuDQzn) entering disabled state
br0: port 1(eth0) entering disabled state
br0: port 1(eth0) entering disabled state
e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
br0: topology change detected, propagating
br0: port 1(eth0) entering forwarding state
br0: topology change detected, propagating
br0: port 2(vethAuDQzn) entering forwarding state

I'm starting to desperate here and i hope one of you has an idea on what
would be needed to make that thing work correctly.

Regards,
Olivier

PS: Sorry if this mail gets duplicated, it doesn't appear to be correctly
sent....
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxcontainers.org/pipermail/lxc-users/attachments/20110415/11c6de4f/attachment.html>