Esxi, HBA pass through and CVM
Posted: Sun Jun 14, 2026 1:45 am
Hi people, what am I missing?
Dual R630 servers with 2 port 12GB Hba's 4 port 10GBe , dual EMC 24 disk SAS enclosure, 2 x 4 port controller's. One server has 2 cables going to one enclosure, 1 cable to controller A and the other cable to controller B. The other server is cabled the same.
Following the deployment guide, two CVM's were deployed and setup with dual Sync and dual iSCSI 10GBe, and the SAS controller's are in pass through mode. Hosts were "added" together, and HA setup. So far, so good.
Now this is where it falls apart. Since each server/array has two connections, all the disks are showing in the cvm as the usual sd* , but when going to create a volume, it states i need to add disk, even though it presents 96 of them.
So I split the systems apart to focus on just one set of hardware, no difference. In the OS of the CVM, I installed and configured multipath and now have the disks showing in the OS a 24 x DM#'s and all physical disks are fully seen with multipath -ll . After a reboot and rescan on the cvm it's still enumerating all the sd# and doesn't show md#'s.
What am I missing?
1. Can't create volume even though there are disks listed.
2. How to handle the dual cable dual sd# per disk (assumed SWCVM would be smart and map WWN to a single device)
3. Now I have the MD's in the OS, but they don't show in CVM
Sort of assume that this all has to be configure via CLI anyway, but man the doco is disjointed. TBH, this setup would be very "vanilla" and the lack of a "simple" setup, i don't get it.
Any thought will be appreciated.
Cheers, Lee
Dual R630 servers with 2 port 12GB Hba's 4 port 10GBe , dual EMC 24 disk SAS enclosure, 2 x 4 port controller's. One server has 2 cables going to one enclosure, 1 cable to controller A and the other cable to controller B. The other server is cabled the same.
Following the deployment guide, two CVM's were deployed and setup with dual Sync and dual iSCSI 10GBe, and the SAS controller's are in pass through mode. Hosts were "added" together, and HA setup. So far, so good.
Now this is where it falls apart. Since each server/array has two connections, all the disks are showing in the cvm as the usual sd* , but when going to create a volume, it states i need to add disk, even though it presents 96 of them.
So I split the systems apart to focus on just one set of hardware, no difference. In the OS of the CVM, I installed and configured multipath and now have the disks showing in the OS a 24 x DM#'s and all physical disks are fully seen with multipath -ll . After a reboot and rescan on the cvm it's still enumerating all the sd# and doesn't show md#'s.
What am I missing?
1. Can't create volume even though there are disks listed.
2. How to handle the dual cable dual sd# per disk (assumed SWCVM would be smart and map WWN to a single device)
3. Now I have the MD's in the OS, but they don't show in CVM
Sort of assume that this all has to be configure via CLI anyway, but man the doco is disjointed. TBH, this setup would be very "vanilla" and the lack of a "simple" setup, i don't get it.
Any thought will be appreciated.
Cheers, Lee