Hello, Harvesters
I have 3 DL360 gen 10 servers all three are identical in settings and hardware top to bottom.
all three are connected to a juniper VXLAN access switch with all three having access ports for management and 3 ethernet connections in LACP trunk for VLAN network. Node1 “cluster manager” is up and running without issues node 2 “worker” is also up without issue and connected to the cluster as a registered node. Node 3 has harvester installed and running this command $ sudo journalctl -b -u rancherd produces the following output that seems to match the same output on node 2
Jul 07 07:16:54 mops-hitl-node3 systemd[1]: Starting Rancher Bootstrap…
Jul 07 07:16:55 mops-hitl-node3 rancherd[2220]: time=“2022-07-07T07:16:55Z” level=info msg="System is already bootstrapped. To force the system to be bootstrapped again run with the --forc>
Jul 07 07:16:55 mops-hitl-node3 systemd[1]: rancherd.service: Succeeded.
Jul 07 07:16:55 mops-hitl-node3 systemd[1]: Finished Rancher Bootstrap.
but this node never shows up in the GUI
any ideas on what or where to look for any indication as to what is going wrong
note node 3: can ping cluster VIP and management of node 1 and node 2
I’ve also wiped the drives on node 3 a couple of times and reinstalled Harvester a couple of times as well all having the same result.
Update-1: after stumbling around the shell I found that under systemctl my worker node has many K3s services running. my non-joined node does not have these services running secondly under rancher2_connection_info.json connect node shows the Kubernetes network address but the non-connected node shows the VIP. Any way of kick-starting the join process from shell, I’ve already reinstalled harvester several times and nothing appears to change.
Update-2: So after re-installing harvester for like the 10th time I decided to change the management IP address used this time around from X.X.X.92 to X.X.X.93 and it joined the cluster after reboot. does anyone here know if the management node stores a record of the previous node or join attempts, odd that changing the IP would matter