Symptoms
Follow VCF on VxRail procedure to deploy VI workload domain cluster.
The VxRail manager default IPv4 address 192.168.10.200 is changed to customer designated address.
Visit this new IP address in browser to open cluster deployment wizard UI, under Resources page, there is no host discovered.
Run command "ifconfig eth1" on VxRail manager, the eth1 has an IPv6 address fe80::xxxxxxxxx
localhost:/home/mystic # ifconfig eth1
eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
inet6 fe80::xxxxxxxxx prefixlen 64 scopeid 0x20<link>
ether 00:0c:29:46:2e:05 txqueuelen 1000 (Ethernet)
RX packets 157 bytes 80631 (78.7 KiB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 125 bytes 22152 (21.6 KiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
Check dayone.log on VxRail manger, do-host query data failed due to it cannot connect to host 9090 port via host IPv6 address fc00:xxxxxxxxx
"2023-05-22 16:30:08,371" microservice.do-host "2023-05-22T16:30:08.141518523Z stderr F 2023-05-22 16:30:08,141 [ERROR] <Dummy-1000:139758144551752> linzhi_dataloader.py fetch_async() (84): Query data failed,url:https://[fc00:xxxxxxxxx]:9090/rest/ps/private/v1/status,Exception:Cannot connect to host fc00:xxxxxxxxx:9090 ssl:<gevent._ssl3.SSLContext object at 0x7f1bfa7f2748> [Connect call failed ('fc00:xxxxxxxxx', 9090, 0, 0)]"
Cause
IPv6 Auto-configuration is enabled on ESXi hosts and TOR switch, so the ESXi host vmk0 is configured with Auto-configuration IPv6 address fc00:xxxxxxxxx
As a result, VxRail manager cannot communicate with host vmk0 fc00:xxxxxxxxx via VxRail manager eth1 link-local address fe80:xxxxxxxxx.
Resolution
VxRail will fix this issue in future release to let VxRail manager only communicate to host vmk0 link-local IPv6 address.
Before the fix is available, please perform below steps as a workaround.
1. Login vSphere host client UI, edit vmk0 and set IPv6 Auto-configuration to "No".
2. Restart loudmouth service on ESXi host
/etc/init.d/loudmouth restart
3. Repeat step1 and step2 on each host.
4. SSH to VxRail manger and restart loudmouth service by below command via root user
service vmware-loudmouth restart
5. Run below command on VxRail manager to check loudmouth output, make sure each host only has one IPv6 address starts with "fe80"
/usr/lib/vmware-loudmouth/bin/loudmouthc query
6. Lauch cluster deployment UI again.
Check below link for details of how to change VxRail manager default IPv4 address before deploying the cluster.
https://docs.vmware.com/en/VMware-Cloud-Foundation/4.5/com.vmware.vcf.vxrail.doc/GUID-D5836F2B-69DA-438C-A34D-23A28E57026F.html#GUID-D5836F2B-69DA-438C-A34D-23A28E57026F