If some services fail to start after an outage and you use systemd, read on.
When a node boots, all its VPSes start one after another, slowing the system down for a while. VPSes using systemd have trouble because services do not finish starting in time. The default timeout is 90 seconds, which is often not enough. systemd considers such a slow-starting service to have failed and terminates it. This affects all sorts of services, including network configuration, databases and web servers.
For this reason, we will add a longer timeout of about 900 seconds to every distribution template using systemd. But that will only help newly created or reinstalled VPSes. Everyone else needs to change it themselves. The setting is DefaultTimeoutStartSec in /etc/systemd/system.conf.
Since this is quite hard to test, please let us know if you still have problems with the longer timeout.