Closed sergiodj closed 4 years ago
Setting a fixed mac address on your bridge usually helps with that.
The Linux kernel somewhat annoyingly picks the highest mac address on a bridge as the bridge's own address.
LXD doesn't do any network changes on instance startup so that's the most likely issue and affects people differently mostly based on luck.
Setting bridge.hwaddr to some fixed value should do the trick.
Thanks, it seemed to work indeed.
Required information
Issue description
When I'm connected to one of my containers/VMs via ssh and I perform a
lxc launch
or alxc copy
&lxc start
, I notice that the ssh connections get closed withConnection to 10.101.133.21 closed by remote host.
orclient_loop: send disconnect: Broken pipe
.The way I work is to have a few "template" containers and VMs, and the copying them over to temporary instances, which means that I'm constantly facing this issue.
Steps to reproduce
lxc launch
or alxc copy
&lxc start
.Information to attach
dmesg
)Not sure this is relevant, but here it is:
lxc info NAME --show-log
)lxc config show NAME --expanded
)This is important. I use a very customized profile.
lxc monitor
while reproducing the issue)