[horde] zend_mm_heap corrupted

[robert sand]

I've seen a lot of discussion about this on php.net and some here too but have not seen a solution for it.

In our case we have 5 horde/imp servers each accessing a single memcache server with 6GB memory dedicated to memcached. 
  I have 2 servers that get these "zend_mm_heap corrupted" messages along with segmentation faults of apache pids.  Yet 
when I direct these 2 servers to a memecache on the localhost it goes away.

Our configuration is:

Debian Lenny:

     * Horde: 3.2.2
     * Imp: H3 (4.2) (run Imp tests)
     * Ingo: H3 (1.2) (run Ingo tests)
     * Kronolith: H3 (2.2) (run Kronolith tests)
     * Mnemo: H3 (2.2)
     * Nag: H3 (2.2)
     * Turba: H3 (2.2.1) (run Turba tests)
     * PHP Version: 5.2.6-1+lenny3

If there is a solution to this could someone point me to it?  This just started showing up with the latest "lenny" 
release of horde.

-- 
Robert Sand.
mailto:rsand at d.umn.edu
1028 Kirby Drive
366 K Plz
Duluth, MN 55812-3095
218-726-6122        fax 218-726-7674

"Walk behind me I may not lead, Walk in front of me I may not follow,
Walk beside me and we walk together"  UTE Tribal proverb.