Safari crashing for network users

[kcalderw]

I recently made an image with Deploystudio for a 10.10 rollout. This was a base image with only core software installed to the local profile with Munki handling most of the other software. I'm seeing no issues with the local account but big issues with our networked users. Users are having a host of issues with Safari like rainbow wheels (having to force quit) and also crashing. How can I correct this? Chrome works fine for everyone, but that's not a solution some like.

 

As someone suggested on a different site, I did rename everyone's Library folder and allow that to rebuild on next login then moved back their Safari bookmarks. This worked for a few days and some users say it sped their computer up a lot. However after a few days, Safari started crashing again. I'm still upgrading our lab to 10.10 (from 10.8.5) so perhaps them going back on machines with 10.8.5 could mess with their library folders but at this point, I'm at a loss.

[furby]

I'm having the exact same problem! Launching the console shows the same errors building up rapidly.

Just now I wiped the last of the 10.8.5 machines and am crossing my fingers but am not hopeful. The next call is to Apple Enterprise.

 

Are you able to load up Mail.app?

[kcalderw]

We do not use the mail.app so I'm not sure. For now I'm testing going back to 10.9.5 for a few users to see if there's a difference.

[furby]

Update (for anyone that might be doing the same). I don't think all Yosemite (10.10) has solved the problem, though things seem a little better.

 

Spotlight seems to be indexing and re indexing at random, especial when using Office 2016 apps. I'm going to downgrade office and then download the El-Capitain Beta and see how that is.

 

Edit/That's odd. Everything pretty much locks up with the log files written to constantly but, if I get a user to log off and back on, everything is fine ¯\_(ツ)_/¯

Edited September 21, 2015 by furby

[norweg]

We had same problem. We used initially "recommended" mcx redirector for "Caches" folder (redirecting it to /tmp/%username%/Caches. Unfortunately how Apple created this feature - does not work well for many processes - as they require access to Caches folder (read/write).

We decided to use loginhook - which creates dynamic link from network "Caches" folder to /Library/Caches/%username%

That seems improved performance and stability...

[kcalderw]

Are you doing that in WGM or Profile Manager?

[norweg]

Actually I use Jamf Casper to deploy package containing script (script to be copied to /Library/LoginHooks), then policy triggers another script to execute "sudo defaults write com.apple.loginwindow LoginHook /Library/LoginHooks/loginhook.sh" once per computer.