Upgrade to 10.9/Server APP 3.0 - Problems with devicemgrd

Question

After upgrading our xserve to 10.9 and Server APP 3.0 the server hangs after some hours, because RAM is full!!!!
The devicemgrd is filling constantly up the RAM and needs about 90% of one of the four cores...
Wiki ist turned off!
What can be the reason for this problem?
Thanks for any help!

Answer 1

Just wanted to follow up on my experience. If you take a look at the thread that I linked to above, you will see the log file and errors listed in it that indicated an issue with the DB data used for the profilemanager. This was exactly the situation I was having, the same error in the same log file indicating the same issue as that thread.
I see that others have been experiencing issues with memory filling up over time as well and those that have applied the same fix I did, but they are still experiencing the same issue. My advice would to investigate the log file listed in that thread and see if you are or ever were having that issue. If not, then that fix won't do much for you.
What I did in order to track down the issue was a combination of Activity Monitor, viewing the memory tab, and using ps ax -v to view which processes were using the most memory. Then I googled issues relating to mavericks memory leaks involving those processes. That ultimately lead me to this thread, and the one that contain the same issue and fix that I needed.
I used Activity Monitor to view the Virtual Memory Usage. I tracked it over time and observed it steadily growing even though no new processes were being started. When I say over time, I mean HOURS worth of time. Also note that Mavericks and most UNIX like OS's will use all available RAM for caching. So the real question isn't if it's using all the RAM, that's normal, the question is does it continue to use climb in usage past the physical memory installed. In my case that is 32GB, my Virtual Memory usage actually rose to 96GB before the system started telling me it needed to pause apps and asking me to force quit. I might add that at this point, I will still able to login via ssh and cleanly reboot the system.
My next step after rebooting was leave all non essential services offline, in my case that was the VMs, plus I wanted to keep from corrupting their filesystems by doing hard shutdowns on them when the host ran out of virtual memory space.
I let the server processes start as normal, then used Activity Monitor and ps ax -v to isolate devicemgrd, google it and here I arrived.
My hope is that you can use some of this trouble shooting to isolate the issues your systems may be having. If you are lucky like me, someone has already isolated the cause and identified the fix. I would have loved to have been the one to come up with the SQL to fix this one, but thankfully I didn't have to experiment on my system, someone already did and was kind enough to share. If you find a fix, please do the same.
Best of luck!