Incident Source: DHCP outage
Affected Nodes: Nodes running CentOS on QRIScloud Stage 2 (Polaris Data Centre)
Incident Description: A 10 minute outage of a DHCP server at Polaris on Friday 5th September resulted in loss of ip address of compute nodes. Debian/Ubuntu VMs were able to reconfigure themselves automatically and regain ip address once DHCP service was restored. Due to a flaw in CentOS the VMs self configured with self assigned ip address and thus did not have internet connectivity.
Clients with password enabled access to the NeCTAR dashboard were able to login and restart the network interface on the affected VMs to restore network access. A restart of the affected VMs also restored network access.