Hello,
I had a general question about a constant error that i'm running into.
I have to Windows 2003 VMs installed on a ESX 3.0.1 server with both VMs sharing a RAID5 datastore that is configured as secondary drive for both VMs.
Everything works fine in terms of access/read/write data, but i consistently get random ntfs error messages in the event log saying the the paritions are corrupt and to run chkdsk utility. Of course i run the utility and but i'll still get the error message upon future reboots for both VMs.
is this an erroneous error message?
or do i need to reconfigure something?
Thanks in advance!
You cannot have two systems (VM's or not) accessing the same disk without some form of locking. The configuration you have has no method of syncing the writes to the disk (like in an OpenVMS situation. Oh, how I miss that OS!)
If you want to connect two Windows VM's (or hardware-based systems) to the same disk, you'll probably need to consider using Microsoft Clusters.
mike
As Mike wrote, you are attempting to use a single system filesystem (NTFS) as a clustered filesystem. You can't do this, and WILL end up with lost data.
You need to find a clustered filesystem if you want to do this, or use a third machine as a network filesystem.
--Matt
I see...and i apologize in advance for I've only been working with the ESX server for 2 months now so just trucking along and finding success through failure...i'm still a newbie.
so then what would be an ideal configuration?
basically, i have a Dell server with one drive dedicated to the ESX server, 2 other scsi drives, each hosting one windows VM and a RAID 5 array for data.
I created a datastore on the RAID 5 array with one VMDK which is being shared between the 2 VMs as its D drive, which alas, the issue with the corrupt disk error...So I thinking, should i just create 2 VMDK on the datastore residing on the RAID 5 array and asign each VMDK to its respective VM?
I appreciate the prompted and informative replies...thanks again.
That would work, yes, though you wouldn't shared data between the VMs....they would each have their own virtual disk.....
--Matt
got it...
thanks again to the both of you...take care.
Another option might be to assign the 😧 drive vmdk file to one of the VMs.
Within that VM, create a network share of the 😧 drive. Then, configure the
other VM to mount that network share as it's 😧 drive.
Just a thought ... 😕
thanks for the suggestion, but either or...it's not a complex necessity so i'll probably go with creating the seperate vmdk(s)...