[horde] more memcached problems

[Kevin Konowalec]

Sorry to bring this up yet again but although everything *seems* to  
be working fine (ie I haven't seen any session timeouts lately) when  
I look in the logs I get thousands of these messages:

Mar 19 13:39:45 src at cluster7 HORDE[24268]: [horde] Error writing  
session data (id = 272083e44eb56eceaa47840a63ee5ad7) [on line 111 of  
"/var/www/horde/lib/Horde/SessionHandler/memcache.php"]
Mar 19 13:39:45 src at cluster4 HORDE[21792]: [imp] Error writing  
session data (id = 4be555aa26e07792d94e878b04514aa2) [on line 111 of  
"/var/www/horde/lib/Horde/SessionHandler/memcache.php"]
Mar 19 13:39:45 src at cluster6 HORDE[32199]: [horde] Error writing  
session data (id = b64e45cdf25d674fb5ed9effd50cb5c4) [on line 111 of  
"/var/www/horde/lib/Horde/SessionHandler/memcache.php"]

Though I could be wrong it appears that these messages appear on  
logout.  I haven't done an exhaustive test on that yet so it could be  
happening at other points in the session as well.

What could be causing this?  I've set up my /horde/conf.php on each  
cluster machine to have the following:

$conf['sessionhandler']['params']['hostspec'] = array('xxx.xxx.xxx. 
187','xxx.xxx.xxx.188','xxx.xxx.xxx.189','xxx.xxx.xxx.190');
$conf['sessionhandler']['params']['port'] = array('3307');
$conf['sessionhandler']['params']['persistent'] = false;
$conf['sessionhandler']['params']['compression'] = false;
$conf['sessionhandler']['params']['lock_dir'] = '/tmp';
$conf['sessionhandler']['type'] = 'memcache';



On each of the 4 machines in the hostspec array I'm running a 1.5GB  
instance of memcached on port 3307:   /usr/local/bin/memcached -u www  
-d -m 1536 -p 3307


Have I STILL done something wrong???  Why would we be getting those  
error messages but yet having Imp work just fine???  How can I test  
the memcached instances to make sure they're writable by the cluster  
machines (there's no firewalls or anything between them so I can't  
see why they wouldn't be...).

Thanks

kevin