Hi Linas, Good to know it. It looks increase the listening time in /etc/init.d/cloud-early-config will fix it (default is 5 times * 2 seconds).
-Wei 2016-12-20 9:23 GMT+01:00 Linas Žilinskas <li...@host1plus.com>: > I don't think the issue is the same. As i mentioned in the original report > and my findings afterwards, this is a specifically qemu issue which was > fixed in 2.4.0-rc3. > > The issue was the way qemu exposes the socket to communicate with VM. It > didn't queue data, so unless the VM was listening on /dev/vport.. at the > time when data is sent, it would never reiceive it. 2.4.0-rc3 fixed this by > queueing the data sent, so once sent, it was accessible (only once) when > the VM checked /dev/vport.. > > On 19/12/16 10:37, Wei ZHOU wrote: > > Hi Linas, > > It seems the issue you mentioned has been fixed by the commits for > https://issues.apache.org/jira/browse/CLOUDSTACK-2823 > > CloudStack-agent will try to pass the boot args 30 times if the console Ip > is not accessible. > > Weird. > -Wei > >
