Today I was finally back at work and off to a customer to install a brand new SCCM 2012 R2 environment for them. Everything was prepared when I arrived and we started off by installing SQL. I had asked for several disks for the installation and had gotten them all, but not as mounted disk, but as mount points on a base drive. This didn’t seem as a problem at first, but at the first disk size check we got an error from the wizard.
We discovered that the SQL-installation wizard weren’t able to view the available space on the mount point and underlying disk itself, only the base drive. In this case our base drive were a mind-blowing 37mb in size. Apparently not enough for a SQL-installation.
As we didn´t fully understood why the mount points were presented in that way, and we were unable to contact the person responsible for the installation of the server and also responsible for the storage, we decided not to assign the mount points one individual drive letter each. This should, I believe, be the first way of solving the issue. But there could of course be other reasons for not wanting to assign drive letters.
After a bit of head scratching and goo.. Oh sorry, Bing:ing we first of all tried to resize the base drive to allow the installation of SQL to proceed. But, unfortunately this wasn’t possible due to some kind of limitation in the Netapp publishing the drives. So, we had to solve it in a more creative way.
– First off all, the mount points was at startup reachable via E:\<mountpoint>. We started off by changing the drive letter of the E:\ drive to S:\.
– Next step – assign one of the mount points the old drive letter, in this case E:\. Now we could easily view the mount point as a drive. On that drive, we created folders with the equivalent names of the mount points.
– Now, during the SQL-installation we pointed out the E:\ drive with the “fake” mount points, and the wizard went all the way through. Note! We first tried to change back to the real mount points at the next step of the wizard, but it does at least one more disc-size check just before kicking of the installation… So we did the entire installation with our “fake” mount points.
– When the installation had finalized we first verified that all services were running, and stopped them. Now, we removed the drive letter assignment from the mount point and changed the base drives drive letter back to E:\.
– The last step is to copy all the installed folders from their “fake” mount point to the real once. When this is done, we restarted the server. And voilà! The services did start and everything looks as it should be.
Note, this is of course a quiet big workaround and I would advise to try the other suggested ways before using this one.