Host has networks that don't exist...

[mlubinski]

Hi,

This is my setup of a cluster (that already has some hosts in it). VC version is Update 4, hosts are 3.5 U4. I will post only config from first (and new) hosts, as I would like to add new host to this cluster, but when trying to reconfigure HA, I got error:

HA agent on new_ESXhost in cluster has an erro Incompatible HA Networks: Host has network(s) that don't exist on cluster members: new_ESXhost_IP: Cluster has network(s) missing on host: first_ESXhost_IP

all current ESX hosts are running fine. Just happens when I try to add new host to this cluster.

Network setup for first host:

vSwitch0:

VM PortGroup

Service Console Port: Service Console, vswif0: IP_Address X.X.158.X

VMkernel Port: VMotion: 192.168.101.1

vSwitch1

VM Portgroups

vSwitch2

VM Portgroups

VMkernel Port: Storage: internal_ip x.x.101.x

-

---

Network setup for new host

vSwitch0:

VM PortGroup

Service Console Port: Service Console, vswif0: IP_Address X.X.152.X

VMkernel Port: VMotion: 192.168.101.6

vSwitch1

VM Portgroups

vSwitch2

VM Portgroups

VMkernel Port: Storage: internal_ip x.x.101.x

-

---

So the only thing that differs between "current" ESX hosts and "new" one is SC IP, that is from different network, but everything is working, when I connect to new esx host, and make vmkping to first ESX VMotionIP, then I get reply, so this should work.

So what to change to make this work?

Thanks in advance for any answer.

[weinstein5]

When the ESX server was set up did you intend for the Service Console port to be on a different segment? It looks like your networking is configured so that you can route to the new management network and since both vmkernel ports are on the same segement then you should be able to vmkping between the two hosts -

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[mlubinski]

I needed to get IP from different network, as this network for currently configured cluster ran out of available IPs. so This is my issue probably, but I should be able to add this host to this cluster, and reenable HA, or not?

[weinstein5]

Can you resolve between the hosts via FQDN? - instead of the vmkping command use the ping command and try to ping via host name - it really should not be a problem to be on a different segment as long as the hosts can resolve by host name -

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[mlubinski]

# ping FIRSTESXHOST.DOMAIN

PING FIRSTESXHOST.DOMAIN (FIRSTHOST_IP) 56(84) bytes of data.

64 bytes from FIRSTESXHOST.DOMAIN (FIRSTHOST_IP): icmp_seq=0 ttl=63 time=0.278 ms

64 bytes from FIRSTESXHOST.DOMAIN (FIRSTHOST_IP): icmp_seq=1 ttl=63 time=0.276 ms

64 bytes from FIRSTESXHOST.DOMAIN (FIRSTHOST_IP): icmp_seq=2 ttl=63 time=0.302 ms

64 bytes from FIRSTESXHOST.DOMAIN (FIRSTHOST_IP): icmp_seq=3 ttl=63 time=0.255 ms

--- FIRSTESXHOST.DOMAIN ping statistics ---

4 packets transmitted, 4 received, 0% packet loss, time 3006ms

rtt min/avg/max/mdev = 0.255/0.277/0.302/0.026 ms, pipe 2

#

[komarm]

Your Service Console on the first, on the second and all others hosts must have the same subnet (because of heartbeat). HA take the first host in the cluster and this is the pattern (backup service console is also important). there is one exception but then you must use advandced HA options called das.bypassNetComptCheck (available VC 2.5 U3 or higher).