Here is the scenario. I have an ESXi 5.0 host running from a USB device on a blade server using shared network storage for its VMs and for it's scratch storage. I have a catastrophic host failure. Can I simply plug the USB device into another blade in the same chasis and boot from that to get my VMs all back up and running.
Also, in a similar scenario if I have two USB devices with the same configuration info on them can I use one as a backup in case of USB device failure or is there config info constantly being written to the USB device that is being used at the time.
At the configuration level, you don't need to, it means that as long as you take regular backup of the config it is ok.
But when you need to setup a backup system if one production system goes wrong and change the blade server, be careful to provide a consistant zoning and LUN masking for that blade, so VMware will see the same LUN and arrays than before. Same thing applies for VLAN, the switchs to which is connected the new blade must be configured the same way the old blade was (its ports must be in the same VLAN), or some VM may not see the network correctly.
1. If a system is corrupted and you just need a fresh USB boot
Plug a fresh USB key in, possibly already installed, boot, then send the archived configuration, reboot
2. If an whole blade goes wrong, you need to change the server.
Put the WWN of its HBA in the same zones as the HBA of the failed blade
Present the WWN of its HBA to the storage array
If you use the same slotin the chassis, networking should be ok, but if not, put the internal ports of the blade switch module corresponding to the new blade in the same VLAN as the internal port used for the failed blade.
If using VirtualFabric / VirtualConnect on 10GBE NIC adapters, you need to configure it the same way it was on the failed system too.
I hope its more clear now !
Hello,
Yes, a key duplicated or moved to another host keeps its configuration, but be aware that this will not properly handle MAC address and WWN zoning (or VLAN tagging for example), so you have to keep some settings consistant outside of the blade.
I think a better way to do this, is to backup the configuration of your ESXs on a regular basis, then when you need a spare, plug a pre-provisonned usb key, then write the desired conf to it and reboot.
Use esxcfg-cfgbackup from a vMA, PowerCLI or vCLI to backup and restore full configs.
Hey there,
So are you saying if I do the regular backups of the config I don't need to worry about the MAC address and WWN zoning stuff or I do still need to worry about it?
If I do still need to worry about it can you expand a bit, I'm completely new to VM.
At the configuration level, you don't need to, it means that as long as you take regular backup of the config it is ok.
But when you need to setup a backup system if one production system goes wrong and change the blade server, be careful to provide a consistant zoning and LUN masking for that blade, so VMware will see the same LUN and arrays than before. Same thing applies for VLAN, the switchs to which is connected the new blade must be configured the same way the old blade was (its ports must be in the same VLAN), or some VM may not see the network correctly.
1. If a system is corrupted and you just need a fresh USB boot
Plug a fresh USB key in, possibly already installed, boot, then send the archived configuration, reboot
2. If an whole blade goes wrong, you need to change the server.
Put the WWN of its HBA in the same zones as the HBA of the failed blade
Present the WWN of its HBA to the storage array
If you use the same slotin the chassis, networking should be ok, but if not, put the internal ports of the blade switch module corresponding to the new blade in the same VLAN as the internal port used for the failed blade.
If using VirtualFabric / VirtualConnect on 10GBE NIC adapters, you need to configure it the same way it was on the failed system too.
I hope its more clear now !
Thank you.
