Re-install a ESXi host when running vSAN
Article ID: 375671
Updated On:
Products
Issue/Introduction
This KB is an informational process of what to consider when re-installing an ESXi host that is part of a vSAN cluster.
- This KB is intended for the scenario of re-installing ESXi for a single host, when the rest of the vSAN cluster is accessible.
*If you have a situation involving more than one host, please contact support. They will be able to provide a detailed action plan after review.
Cause
An ESXi host used in a vSAN cluster needs to have the OS (ESXi) re-installed without impacting vSAN cluster'
Resolution
ESXi may be reinstalled on a vSAN host so long as the vSAN disks remain untouched throughout the process and the host starts in a healthy state.
- Be sure to take note of the configuration prior (such as IP addresses, vLANs and the Subnet for vSAN, vMotion, and Management VMKernel adapters. As well as which drives are in use and have vSAN data on it).
- See caveats in 'Additional Information', which will cover the considerations for the vSAN host when re-installing ESXi.
Instructions: before you re-install ESXi, if these steps are possible, please perform them to ensure the host will come back smoothly. [Actions are on vCenter]
- Place ESXi host into maintenance mode.
- Right click the host 'connection', and click 'disconnect'
- Right click the host 'Remove from inventory'
Instructions: post re-install of ESXi:
Configure management network for host,1 through DCUI.
- https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.vsan-planning.doc/GUID-8105EE33-2346-4299-9540-41C488A2A8D0.html
- Add host into the 'Datacenter' folder in vSphere (do not add to the vSAN cluster directly, this is crucial) and DO NOT use vSAN quick start; Ignore it. Using this on a host with an existing cluster could result in data loss.
- Configure the vSAN networking, vMotion, and update to management for the host.
- Test the connectivity of each VMK (Management to Management, vSAN to vSAN): https://knowledge.broadcom.com/external/article/344313/testing-vmkernel-network-connectivity-wi.html
- After the host is configured and network is tested. From vCenter, drag and drop host into the vSAN cluster (this will push vSAN configuration update)
Side note: when re-installing be sure to pick the drive intended for ESXi OS (as vSAN drives may appear as an option for install when selecting install location).
Additional Information
1: vSAN object health.
The Host should be in maintenance mode (place in 'full evacuation' if possible; other wise 'ensure accessibility' will work, where rebuild of objects will start after an hour by default).
Alternatively if the host is offline, check the health from vCenter or another host.
- The objects should not be inaccessible prior (reduced availability or healthy).
- If there are inaccessible objects, then this means the cluster is unhealthy data wise, where taking out another host or re-install could return to this state or make the situation worse.
Check the vSAN object health from the host using CLI (any host).
- esxcli vsan debug object health summary get
2: Encryption.
- If encryption is used, the KMS must be online and accessible in order to decrypt the vSAN drives after reinstall.
- If the drives are locked when the host is added back to the vSAN cluster (post configuration).
- Try to reboot the host once in the vSAN cluster (as keys are sent and decrypted on power up).
3: Networking.
- The host needs to be dis-associated from the VDS if they were a part of one (If gracefully removed from inventory, it should do this automatically).
- You can re-add to VDS after re-install and before adding to vSAN cluster.
- After the re-install, you will have to manually setup management network in DCUI. Afterwards, you can setup a vSAN vmkernel port in the Host GUI for standard or vCenter for VDS, in order to communicate to vSAN.
*Be sure that the IP addresses, vLANs and Subnets are recorded to ensure they can be re-set as they were before.
4: vSphere/vCenter.
- The host should NOT be in vSphere cluster at time of re-install (as it will need to be re-added from datacenter folder, after configuration. This will ensure it pushes vSAN config to existing cluster).
- Do this by doing a right click on the host, under connection, choose 'Disconnect'. Then right click the host again, and do 'Remove from inventory'
5: Certificates:
- If the host had a custom certificates with vCenter in Custom Mode (meaning the ESXi Machine Cert was replaced). You will have to temporary set it to 'VMCA' mode to allow host into inventory. Where you can replace the ESXi certificate and change the certificate mode back to custom.
Feedback
