Hi community
I wanted to know what your experience was of rebooting nfs storage devices WITHOUT rebooting the ESX servers. I can shutdown the VMs that use the storage but would like to avoid shutting down the ESX hosts as they require a visit to the data centre to switch on and I can do all the rest from home.
My options are to either then reboot the ESX hosts (which I can do remotely of course) or will the ESX server automatically reconnect to the nfs storage when it is back up?
This is the first time such a reboot will be necessary so I am using
this as a learning experience, and the vms on these ESX hosts are not
critical. (although don't tell the users that. )
thanks in advance
Yes, of course you can.
ESXes will automatically reconnect to the storage.
---
MCSA, MCTS Hyper-V, VCP 3/4, VMware vExpert
Yes, of course you can.
ESXes will automatically reconnect to the storage.
---
MCSA, MCTS Hyper-V, VCP 3/4, VMware vExpert
...and you do not even need to power off your VMs.
vcenter event log when the storage was rebooted:
Message on gvc-002 on vhost-014.myplace.com in CampusDC: NVRAM: write failed info 23/06/2010 08:02:11 User
and when it was available again:
Restored connection to server 192.168.19.243 mount point /vol/vmstore02 mounted as eb2a419e-62e30we0-0000-000000000000 (vmstore02). info 23/06/2010 08:13:52
and the OS of the VMs, both linux and windows, just hung until the storage was available again. No errors were logged in the VM OS at all.
great!
I know this question has been answered, but I thought this follow up could be useful for other readers who might be wondering about the same thing.
It seems like NFS / NFS handles this very well. We had a couple of unfortunate occations where we lost connection between a group of ESXi 4.1 hosts and their NetApp NFS Server. Once we lost connection for apx 5 mins and experienced no other issues than a few Windows event viewer errors (disk timeoust). The second horrible occation, which resulted in apx 30 min loss of connectivity, caused a few of the servers to reboot, but about 95% still managed with only event viewer warnings and error messages. Although I wouldn't recommend trying it in a production environment, we haven't expereienced any corrupted databases or data loss from these situations. Now we just need to get rid of thos d..n Nortel blade switches