I have a PE 2950 and a PE 2650 (and a PE 2550, decommissioned). My 2950 is having issues with Exchange; I didn't follow all best practices when setting it up (I'm beginning to wonder now whether I followed any), so I'm trying to remove other software that could be conflicting. We have our tape drive and PV220S hooked up to this server (PERC 4e/DC). The 2550 has a PERC 3/DC, which I think I can move to the 2650, then I can move the tape drive.
I'm wondering what I should do with the PV, though. It has to be accessible from the 2650 to back up the contents to tape. If I move it over to the 2650, will it keep virtual drives intact and will the new server recognize them? (I know when I set them up, I had to initialize them from Disk Management, and if I had to do that again on the new server - I'm thinking I would - the data would be inaccessible.) Would it be possible to switch from joined bus to cluster and just attach both servers? (It was on cluster originally, but not used that way, so I don't know what configuration, etc. I'd need to do... when I switched it to joined bus, I couldn't add the last disk to the recognized disk space in Disk Management without deleting and recreating the partition, which we didn't want to do at the time, so I don't know if it's in use by the PV/RAID, or not.)
Hopefully this is understandable. If you need clarification or further information, please ask.
Just to make sure I understand what you're saying:
- you have the PE2950 using a PERC4e/DC connected to the PV220S on one side
- you have another Poweredge using a PERC3/DC connected to the other end of the PV220S
But you're not running a cluster? (usually a cluster has 2 identical servers using identical raid controllers)
This would be why your exchange is having issues; you can't share disk space on a block level device (e.g. SATA/ATA/SCSI/SAS/iSCSI/fibre channel) without running an actual cluster (e.g. Microsoft Cluster services).
Reason/example why/what will happen:
- one server sees some freespace and wants to write a file to that free space, so it does
- the other server doesn't know the first server is writing to the space and wants to write to the disk (even if just as simple as updating the archive bit after a file has been backed up (to tape)) and writes some data to the 'free space'.
Now you have a corrupted filesystem.
A cluster prevents this by only allowing 1 server at a time to write to (and read from) the disk at any time.
I suggest you:
- find a PCIe SCSI card and put it in your 2950 to connect the tape drive/library to
- disconnect the other (non-2950) server from the PV220S
- power off the 2950, power off the PV220S, change the PV220S to single bus mode, power on the PV220S, power on the 2950
Post attempt #3
Member since 2003
I'm sorry I didn't respond. I thought I would get an e-mail update when there was another post, but apparently not...
Actually, I had recently turned 'cluster mode' off, to get the benefit of the last disk, because we only had/have the PV220S attached to one server (the PE2950)... soon after I flipped the switch to joined bus, I had this issue, so I *wanted* to connect a second server and actually use cluster mode (as I said, before, it was set to cluster mode, but we were not actually using it in a cluster).
I worked around my original issue by installing the tape drive/backup software on a server with more diskspace. Now I replicate from the PV220S to that server, and then back up to tape.
Thank you for your explanation... I do understand clustering a little better now (although now I'm not planning to use it :).