On Fri, 23 Sep 2011, Stephen Gallagher wrote:
This isn't enough information to say where the latency is occurring. Is the 30s passing while it's still iterating through "Looking up parent groups for group", or are you seeing it happen while the SSSD is saving the results to the cache? I'm guessing that the problem is likely happening during the latter. We're doing a lot of work right now for 1.6.2 with improving the cache performance, but I'd like to try and track this down to make sure our current enhancement plans will aid you.
This was definitely one of the cases with 1.6.1 that I found was cache performance limited (as you suggest). In my case symlinking cache_default.ldb to /dev/shm made this clear by entirely resolving the performance issue (reducing a near 60 second op to around 3).
jh