Greetings!
I am replicating my production environment in a virtual server. The host is a Sun x2200, single quad-core CPU, 8GB RAM, and a QLogic 2460 FC adapter to my SAN. This is running the 64 bit version of Red Hat Linux 5.3
On this host I have created 6 guests. In order to more precisely re-create my production environment, I mapped the LUNs from the production boxes (after powering them off) and new, temporary LUNs to the host and used the linux command dd to replicate the data, networking, etc. I created virtual servers an used the LUNs as scsi passthrough devices to get to the data. In all I am running 16 LUNs in this way.
My guests are as follows:
Windows 2003 Server
NetWare 6.5.7 doing file and print services
NetWare 6.5.7 running GroupWise (this is being migrated to OES)
SuSE Enterprise 10 SP2 also running OES 2 (this is going to be the new GroupWise server)
NetWare 6.5.7
Windows XP Pro management workstation
In the test environment all systems' network cards are set to host-only so I can run it without impacting my production environment.
Everything works except for when I stress the data path on my email server. In order to migrate the data from the current mail server to the new one I have to copy all 30 GBbetween the machines. I have tried several methods to make it happen and all have failed.
1. I can mount the data volume from the NW server on the OES server and copy the data.
2. I can re-map the LUN to the OES server and copy it directly from folder to folder.
3. I can map drives on the management workstation to both servers and copy it that way.
Every time, about 3 GB into the copy I get a MASSIVE number of errors on the guest (be it the Netware or the OES guest) and the NSS volume dismounts. No errors show up at all on the host or on the SAN. Here are some of the errors I get on the guest:
Device not ready: <6>: Current: sense key: Not REady
end_request: I/O error, dev sdd, sector 98497911
NSSLOG ==> zlssLogicalVolume.c[4989]
Error reading PoolData Block 5241990, status=20206
Pool "MAIL" - MSAP deactivate.
I have tried re-creating the volume several times with no additional success.
My question is this: Is it possible that VMWare is having trouble with 16 passthrough devices? The LUN in question has been #13 and #16. Without a bunch of work I can't move it easily to #3.
Since the issue shows up on multiple guests with different OS's, with multiple LUNs, and only when large amounts of data are being transmitted, I have to think it is either a HBA driver issue (it is at the latest version) or a VMWare issue.
Anyone else ever done something like this? I need to finish this test so I can do the live migration on Saturday.
Any help would be GREATLY appreciated.
Thanx!
-Michael