Community

Important Notes:

• There are 2 important site.conf variables that can't be changed after initial seed installation, please be sure the the following configuration variables are set correctly:

1. The time (The time zone must be set correctly, and we recommend specifying NTP servers in site.conf for accurate time synchronization)
2. The cluster subnet (The “[[[ X.Y.Z.0/AB ]]]” variable under “[[clusters]]”)

Other site.conf variables can be changed later. (see nimbula-admin get configpoint <key>)

• The seed node is temporary, doesn't install anything on the local disk, and needs to be brought down once the installation of the 3 first nodes is completed, and the API is available

• There should be no DHCP server on the Nimbula Director dedicated native VLAN

• Nimbula Director does NOT support multiple network interfaces of the same server being linked to 2 different VLANs or layer 2 broadcast domains

• Installation on bare metal is required when using Nimbula Director with KVM (Nimbula Director runs in a VM in case of ESXi 5)

Before submitting a support ticket, please look through the following checklist to ensure your setup conforms to a minimal Nimbula Director platform specification:

1. Minimum number of servers: 4 or more to have HA, 3 minimum without HA
2. CPU bus width: 64bit
3. CPU: VT-x or AMD-V Extension Present
4. BIOS setup: Intel VT-x or AMD-V enabled on all servers
5. Number of cores per server: 4 minimum
6. RAM per server: 8GB required minimum, 16GB or more recommended
7. Minimum amount of local disk space (If using RAID, minimum size of block device presented to operating system) 200GB required, 500GB preferred, 1TB optimal
8. BIOS setup: HPET enabled
9. BIOS setup: Boot order - disk before network on all servers
10. BIOS setup: AHCI enabled if using SATA hard drives
11. Network adapter: PXE capable
12. Network adapter: PXE ROM enabled in BIOS setup
13. Network adapter: PXE ROM enabled on NIC linked to switch
14. Network: No DHCP server on Nimbula Director dedicated native VLAN or layer 2 broadcast domain
15. Network interface for all servers: Linked to same native VLAN or layer 2 broadcast domain for a single cluster site.
16. Network cable or link per server: Either un-plugged or plugged in to same native VLAN or layer 2 broadcast domain
17. Network connectivity per server: Is NOT linked to 2 or more different native VLANs or layer 2 broadcast domains (Nimbula Director does NOT support multiple network interfaces of the same server being linked to 2 different native VLANs or layer 2 broadcast domains)
18. Switch configuration: LACP disabled (can be enabled if 802.3ad is set in site.conf)
19. Switch configuration: STP enabled
20. Switch configuration: Port linked to servers configured with "port fast" (Cisco), or "edge" (Juniper) options.
21. DNS servers: Respond to requests within cluster and instance subnets
22. Router configuration: Default gateway responds to ping
23. Router configuration: Netmask is 1 less than cluster and instance subnets. If cluster and instance subnet are /20, router should be configured with /19 netmask to route traffic from cluster and instance subnets. Example: Network 172.16.0.0/19, Default gateway: 172.16.0.1/19. If Cluster is 172.16.0.0/20 and instance is 172.16.16.0/20.
24. Router configuration: Linked to same native VLAN or layer 2 broadcast domain than server nodes running Nimbula Director software
25. Cluster subnet: Should be including default route IP address. For example, if default route is configured with 172.16.0.1/19, cluster network should be 172.16.0.0/20.
26. Cluster subnet: Netmask equal or above /20 (4094 addresses)
27. Cluster subnet: No network device should be configured with an IP address belonging to the cluster subnet, other than the default route. Nimbula Director assumes the entire subnet is available, except for the default route. Special site.conf parameters exist to specify cluster DHCP IP ranges in case the cluster subnet is not fully dedicated.
28. Instance subnet: If router is configured with default gateway 172.16.0.1/19, then instance network should be 172.16.16.0/20
29. Instance subnet: Netmask equal or above /20 (4094 addresses)
30. Instance subnet: No network device should be configured with an IP address belonging to the instance subnet. Nimbula Director assumes the entire subnet is available. The default route used for instance traffic will be the same as the one used for cluster traffic.

Edited jeremy (Dec. 7, 2012 13:47:22)