Spotlight on a Reshare

[JonThompson]

So, I noticed that I was getting errors on my reshares related to spotlight. Turns out, the check boxes on all of my XSan reshare volumes are grayed out.

The question is: What good is spotlight support and all of the business file support that Apple touted if I can't index a reshare? I have spotlight turned on for the XSan Volume, but it won't let me turn on the individual indexes. You would think that Apple could parse out the San's spotlight index to just include a reshare.

Anyhow, if this is a bug, not a feature, please let me know so I can rack my brain as to how to solve it. In the meantime, is there any way to get rid of the mds errors that keep coming up for my reshares?

[flyingcactus]

I was having the same issue. It turns out there is a KB article on apples website that corrected this issue for me. (http://support.apple.com/kb/HT1139?locale=en_US) It said:

[rstasel]

by clients, do you mean afp clients?

[flyingcactus]

Yes the AFP clients

[r520m]

Anybody got a healthy, fast, reliable Spotlight after the volume failover?
I'm resharing the Xsan volume on two Xsan clients and it seems the Spotlight search is slowly dying over time.

[morphenine]

I just went through a similar thing with apple enterprise support last week.
I had been having major problems with spotlight and MDS errors. I was also having race conditions to develop in the .Spotlight-V100 folder when AFP users connected, and this would eventually deteriorate into "invalide inode" errors.

After 3 open cases and several hours on the phone I finally reached a senior xsan engineer, who basically told me that the spotlight feature in Xsan 2 "doesn't behave the way that we'd like it to, nor the way it was intended." (but nevertheless the same marketing people who sold us "MultiSAN" got ahold of this and ran with it). For instance it is known that after a failover, the spotlight feature will not work on the other controller (I believe this has something to do with the machine's GUID in the index, but I'm not entirely clear on it myself).

In my case, I had followed the KB article 4 times to rebuild my index, but it just never stayed working for more than a couple hours (http://support.apple.com/kb/HT1134). After working with him for several days, the engineer told me that in order to solve my issue, I would likely have to disable spotlight. Well my issue was solved by turning off spotlight on the volume settings, doing a cvfsck, and deleting the .Spotlight-V100 hidden folder from the volume. I no longer get MDS errors, nor invalid inode errors and everything works great...exept for the fact that spotlight is off, but in my environment thats no big deal.

Another thing you might try (that the xsan engineer suggested) is on your clients (xsan clients, and afp clients) to add the volume to the spotlight privacy list. It looked like I also had some machines that were trying to index the volume via afp and causing other problems.


Other Thoughts: When asked about a maintenance release (2.1.2 or 2.2) he seemed hesitant, and without saying anything directly, it sounded like he was hinting that this "feature" wouldn't be fixed until xsan3 at the earliest...but thats just what I read into it.

[abstractrude]

I finally reached a senior xsan engineer, who basically told me that the spotlight feature in Xsan 2 "doesn't behave the way that we'd like it to, nor the way it was intended."


BINGO. Apple can suck it on these half ass implementation of new features. I would have bought xsan2 for more stability, why waste our time with new broken code?

[JonThompson]

[r520m]

Very interesting.
I made a test on another Xsan volume with only one MDC and forced the MDC to shutdown using the power button!. After starting it up again I had the following in the system.log:
Apr 20 11:55:12 mdc1 mds[64]: (/.Spotlight-V100/Store-V1/Stores/55D3D56C-789C-452E-80E5-A41E9F3D4A70)(Error) IndexCI in ContentIndexOpenBulk:Unclean shutdown of /.Spotlight-V100/Store-V1/Stores/55D3D56C-789C-452E-80E5-A41E9F3D4A70/live.0.; needs recovery

Spotlight is working und updating the index!
So it looks like the failover from one MDC to another is rendering the Spotlight index unusable. I will mount the Xsan volume on MDC2 read-only and see if this makes a difference to Spotlight when the volume is back on MDC1.

[pnerum]

I'm curious if there are more successful spotlight enabled reshares.
Up till now I didn't enable Spotlight on my SAN volume, but my users are "almost" screaming for it....

Is this the way to enable it? http://support.apple.com/kb/HT1134

With kind regards,

Patrick

[cedge318]

That's the way we've gone about it. In my experience either spotlight works or it doesn't. There's been no in between...

[wrstuden]

[CitraMalus]

Dont know whether my reply would be of great help.. but have you guys tried out erasing the index and recreate it.?

if not ... then try that once and turn ON the indexing in the specific volumes.

Hope it may work...

[mactinosx]

Hi,

I had the same Spotlight problem on my client "re-share" server (this is the server what is re-sharing the Volume to all our network clients). I did the following thing that worked for me;

- Unmount every volume (mdc2 fail-over and the client server) in the Xsan but keep the volume mounted on the mdc host server (mdc1).
- Delete the com.apple.spotlight.plist in your preferences
- go to the Terminal and login as root and re-index Spotlight by typing the following command:

[endast]

Hi,
Did you delete the com.apple.spotlight.plist on the mdc host server or the client (or perhaps both)?