We are having persistent spurious "xen-boot" failures. Cursory investigation revealed that this was due to problems with NTP startup, so we put it down to infrastructure problems.
However, it turns out that the cause is that we are using the Debian-default NTP servers from pool.ntp.org. We might expect some of these servers don't work some of the time. The underlying cause is a bug in Debian: the NTP servers specified via preseeding are used during the installer environment to set the clock, but the information is not used to configure the actual ntp server package in the install. I have reported this as a bug in Debian but we need to work around it by editing ntp.conf after installation. There are also two unrelated fixes in this series. One of the changes (`Provide get_target_property') is likely to be useful for nested virt. In the absence of objections I am going to push this series soon. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
