Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have got 2 x HP DL360 G7 servers with the sc08 controller cards installed, and dual controller P2000 G3, the cabling is a single cable from each server to the controlelr cards (using SAS port 1 on each controller).

I have got the disk array all setup and the vdisk is visable to each server as a 3TB volume, I formatted it on server1 and it was assigned the drive letter E.

I rescanned for disks on server 2 and it found the formatted volume and put it straight online as E, so I figured I was onto a winner.

However it appears that while both servers see 3TB they can't see each others data, is there a way for them to see the same data (be it additional cabling / tick box in the config somewhere)

thanks in advance.

share|improve this question
up vote 2 down vote accepted

You need to use a cluster-aware file system - plain vanilla NTFS, that I'm pretty sure you're using, isn't one.

The issue is that whenever two boxes share the same data something needs to manage the locking so that one server doesn't overwrite another's data and cause corruption.

In a NAS environment the NAS protocols themselves (say NFS, CIFS/SMB etc.) and the NAS controllers inherently manage this locking process. In a SAN environment (FC/FCoE, iSCSI) the servers themselves have to manage this process, meaning it's actually the file system on those boxes that do this job.

Now I have to tell you right away that what you've done already will have very significantly corrupted the volumes you've already made, I have to say that so you can understand the impact that has on your project and timelines.

Now you need to figure out a way forward by selecting a cluster-aware filesystem. There are a number of these, in a VMWare world their default filesystem (VMFS) is cluster aware, as is Oracle's OCFS/OCFS2 in their linux world and there are others in the linux space too.

For Windows there's two options I'm aware of, the one I use is Symantec's 'Veritas Storage Foundation for Windows' - not only does it allow you to create an NTFS-compliant clustered file system but it'll also add in some multi-pathing options too - that's why I use it, it's far from free however.

What you can also use is MS's own cluster service, this can be a bit daunting at first to configure but essentially it adds a new service to your Windows servers that handles the block-level locking that's needed, you'll have to implement the multipathing separately using MS's MPIO mechanism but this scenario is at least included in most versions of Windows Server. In fact this is the recommended method for setting up older versions of MSSQL, so there's lots of documentation around to help with that.

So to summarise, this stuff doesn't just work on its own, you have definitely killed your file system, hopefully that's not an issue - and you need to pick a way forward and deploy that.

share|improve this answer
At the moment the total data loss is 2 folders with nothing in them so all is well at this point. With MS Server failover clustering (using W2k8 R2 on each host) Is there an option to run stuff on only 2 host as all I am really after is the shared storage bit of it, I'm happy to just have print services on 1 server and the management software on 1 server, oddly its the filestore that needs high availability. – Liquidkristal Jul 20 '12 at 12:51
You may need to carve that storage up a little bit more, IE... make one volume a dedicated volume for the server running the printing role, and another volume for whatever else you're doing. During normal operation role A stays with server A, and role B with server B, but they'll always have the option of failing over/running off one box and when they do so, the cluster volume necessary to run the services goes to the necessary cluster node. – SpacemanSpiff Jul 20 '12 at 12:53
got ya. each server has 300GB of local storage, would using that for the stuff I am not really bothered about having HA for be feasable? – Liquidkristal Jul 20 '12 at 12:56

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.