I’m wondering if there is a best practices recommendation for Spotlight indexing on Drobo units connected to a Mac OS X system? Has anyone looked at performance issues leaving indexing turned on for the entire Drobo volume or set of volumes vs. turning it off completely vs. only partially indexing specific folders/directories? I didn’t find any previous history on this searching threads for “spotlight” so apologies is this has been covered somewhere else - I didn’t find it.
I’m not familiar with Spotlight in particular, but Drobo definitely slows down during simultaneous access.
This means you don’t want indexing to happen when you’re “doing” stuff.
And when you’re not “doing” stuff, you want the indexer to index only one volume at a time on a Drobo, as the Drobo is a single physical volume.
I had an issue with my DroboPro never spinning down the drives after 15 minutes of inactivity. Found that spotlight was the cause - disabled spotlight and fixed the issue.
In terms of performance if you are using half-duplex connections like USB2 or Firewire 400 then you probably want to disable spotlight as it may cause a noticeable performance impact if you create or update a lot of files.
BTW you can also mount the filesystem with the ‘noatime’ option in OS X or Linux to further improve performance by cutting down on unnecessary writes to update file access times.
I have Spotlight turned on on my whole Drobo, all 2+ TB. I don’t think it’s causing an issue for me.
It has been on since the beginning as the data size has grown from zero to where it is now, so I can’t comment on how it would behave if it had been off and was then turned on against a large volume for the first time. I don’t see it ever sat indexing for extended periods of time, even after a reboot.
I don’t have Time Machine backups on there at the moment, I had an external drive to do that before I got the Drobo and haven’t changed my scheme, so it’s mainly fairly big media files (TV, movies, lossless audio, photos, etc.). This use case is probably a fairly easy one, as it can really only index the file names and a bit of metadata, the file contents are not likely to be indexed like they would be if it was full of documents.