VMWARE ESXI 6.5 LSI LOGIC PARALLEL SCSI ADDITIONAL ADAPTORS and Failed - VMware ESX cannot open the virtual disk

[Howzit42]

Hi

After some problematic attempts I managed to configure two LSI LOGIC Parallel SCSI Adaptors using the browser client. I have 11 thick provisioned drives and I assigned 5 to the one adaptor and 6 to the other. Can one have more than two LSI LOGIC Parallel SCSI Adaptors? If I want to configure four adaptors is this possible or do I need to select another adaptor type such as LSI SAS SCSI Adaptor?

When I try and power up the VM I get an error message Failed - VMware ESX cannot open the virtual disk "/vmfs/volumes/5b8c88c5-fea7e846-9d00-c81f66b90358/apbco-vm-092/apbco-vm-092_9.vmdk" for clustering. Verify that the virtual disk was created using the thick option. There are no snapshots of this VM.

I configured all the drives on the first SCSI adaptor and removed the second SCSI adaptor but I still get the above error when powering up the VM.

Is there a way to repair these disks using without losing the data?

Any help is appreciated. Thanks.

[a_p_]

Has this been working before, or is it a new VM (cluster)?

You can basically have up to 4 virtual SCSI controllers. Depending on what you are trying to achieve, clustered disks may need to be connected to a secondary (i.e. not the first one) controller, and virtual disks have the be "Eager Zeroed Thick".

If you can't get this to work, then please attach (don't paste) the VM's current vmware.log to your next reply.

André

[Howzit42]

Thank you for responding. It was a working VM.

It is a standalone VM loaded with Centos 7 64bit OS.

It is not in a cluster nor are any of the disks cluster disks.

I wanted to add SCSI controllers to increase throughput.

All the drives are thick provisioned. Hard disk 3 is Eagerly Zeroed, the rest are all Lazily Zeroed.

All of the drives contain data and were working prior to adding a second SCSI adaptor.

[a_p_]

Can you please clarify some things.

You mentioned that you've configured two SCSI controllers and distributed the virtual disks across them, but the log file only shows a single controller.

The error messages in the log file point to a configuration issue, where multiple VMs use the same .vmdk files!? Can you confirm that there's only one VM which has the virtual disks connected?

André

[Howzit42]

Hi Andre

As per my original post

"I configured all the drives on the first SCSI adaptor and removed the second SCSI adaptor but I still get the above error when powering up the VM"

so the vmware logs will only show one SCSI controller.

I have managed to get the VM up again by running vmkfstools --eagerzero on every drive.

There is only one VM that has these drives connected which is apbco-vm-092.

There are two other VMs confiugured on this host, namely an IPFire firewall which is running and a backup of apbco-vm-092 named apbco-vm-092bu

which is powered off. apbco-vm-092bu also only has one drive on which I re-installed Centos 7.

The problems I experience when trying to add a second SCSI adaptor is an error message This location is being used by another device, taking next available location.

It does not prompt me or allow me to specify controller 1 when adding the second controller. If I change SCSI Bus Sharing to Virtual on both controllers and save the

configuration I get an error in the main window A specified parameter was not correct: unitNumber and the additonal SCSI adaptor is discarded from the configuration.

So in summary, I have sorted the can't boot device drive issue running vmkfstools --eagerzero on every drive but I still have been unable to add a second SCSI adaptor,

with exception of one occasion which I have been unable to  replicate.

Regards

Pete

[a_p_]

Sorry that I've missed the part where you reverted to a single controller.

I'd like to see whether I can reproduce (and maybe workaround/solve) this in my test environment. Would you mind to attach the VM's current configuration (.vmx) file to a reply post?

Also, which client do you use the configure the VM? Is it the Embedded Host Client on the ESXi host itself (likely v1.27), or is this host managed by vCenter Server, and you are using one of the web clients from vCenter Server?

André

[VMSysProg]

Just confirming, on original question, I do have VMs (RHEL 7.9) running either FOUR VMware Paravirtual scsi controllers, or FOUR LSI Logic Parallel SCSI controllers with no issues.   Both work for Oracle workloads:

SCSI 0 -- all RHEL file systems (/, swap, /u01 (Oracle), /u01temp (Backups from FRA to disk, temp files)
SCSI 1 -- all VMDKs mapped with udev to /dev/asm-DATA-... for Oracle Grid/ASM to make disk group for database
SCSI 2 -- all VMDKs mapped with udev to /dev/asm-FRA-... for Oracle Grid/ASM to make disk group for FRA / archive / recovery 
SCSI 3 -- all VMDKs mapped with udev to /dev/asm-REDO-... for Oracle Grid/ASM to make disk group for REDO Logs

Note:  udev (/etc/udev/rules.d/99-oracle-asm-devices.rules) is used to map SCSI id to stable device names that Oracle ASM can see since RHEL kernel does not guarantee order of SCSI controllers at boot time which affects /dev/sd[a-m] letter is assigned.  The udev rules file allows me to to map the SCSI ID (UUID) to the device name I want in /dev

For instance:

[oltpmain:larryt:nodb:~]$ ls -l1 /dev/asm*
lrwxrwxrwx. 1 root root 3 Oct 16 13:37 /dev/asm-ORA_SE2_DATA-OM-P-1 -> sdd
lrwxrwxrwx. 1 root root 3 Oct 16 13:39 /dev/asm-ORA_SE2_DATA-OM-P-2 -> sde
lrwxrwxrwx. 1 root root 4 Oct 16 13:39 /dev/asm-ORA_SE2_FRA-OM-P-1 -> sdg1
lrwxrwxrwx. 1 root root 4 Oct 16 13:31 /dev/asm-ORA_SE2_FRA-OM-P-2 -> sdh1
lrwxrwxrwx. 1 root root 4 Oct 16 13:39 /dev/asm-ORA_SE2_REDO-OM-P-1 -> sdi1
lrwxrwxrwx. 1 root root 4 Oct 16 13:31 /dev/asm-ORA_SE2_REDO-OM-P-2 -> sdj1

is mapped by 

[oltpmain:larryt:nodb:~]$ cat /etc/udev/rules.d/99-oracle-asmdevices.rules
# sdd1 DATA 1 270GB
ACTION=="add|change", ENV{ID_SERIAL}=="36000c29b58136cf693eb2f36bb7fcc6b", SYMLINK+="asm-ORA_SE2_DATA-OM-P-1", GROUP="dba", OWNER="oracle", MODE="0660"

# sde1 DATA 2 270GB
ACTION=="add|change", ENV{ID_SERIAL}=="36000c293af59911daeb650174c6cf29d", SYMLINK+="asm-ORA_SE2_DATA-OM-P-2", GROUP="dba", OWNER="oracle", MODE="0660"

###

# sdg1 FRA 1 300GB
ACTION=="add|change", ENV{ID_SERIAL}=="36000c29343f387359b53028b5e513964", SYMLINK+="asm-ORA_SE2_FRA-OM-P-1", GROUP="dba", OWNER="oracle", MODE="0660"

# sdh1 FRA 2 300GB
ACTION=="add|change", ENV{ID_SERIAL}=="36000c29e64f708f5b617f80569ad7522", SYMLINK+="asm-ORA_SE2_FRA-OM-P-2", GROUP="dba", OWNER="oracle", MODE="0660"

###

# sdi1 REDO 1 40GB
ACTION=="add|change", ENV{ID_SERIAL}=="36000c2954d99e1bf1c7a2b67bb995a51", SYMLINK+="asm-ORA_SE2_REDO-OM-P-1", GROUP="dba", OWNER="oracle", MODE="0660"

# sdj1 REDO 2 40GB
ACTION=="add|change", ENV{ID_SERIAL}=="36000c29e27c1793c575389737b13cf90", SYMLINK+="asm-ORA_SE2_REDO-OM-P-2", GROUP="dba", OWNER="oracle", MODE="0660"

[oltpmain:larryt:nodb:~]$

[VMSysProg]

FYI -- one challenge I do have is to determine if there is any way to change LSI LOGIC PARALLEL TO VMWARE ParaVirtual SCSI controllers without destroying data integrity of the 10 drives mapped (8 for Oracle, 4 for RHEL file systems and swap).