[horde] memcached errors...

[Kevin Konowalec]

Haven't gotten any responses to this so I'll just give this one a  
bump...


Horde: 3.1.2
Imp: H3 (4.1.2)

We've just switched over to using memcached sessions.  It's SMOKING  
fast and working well so far but in my logs I'm seeing a LOT of
"Error retrieving session data" errors.  ie:

Aug 29 13:42:32 webcluster10 HORDE[510]: [horde] Error retrieving  
session data (id = c8abe116058606a6acc1cf76d97fd769) [on line 83 of "/ 
var/www/horde/lib/Horde/SessionHandler/memcache.php"]

We're running a 12 machine cluster with PF round robining.  Users   
only have the chance of being redirected to a different server if  
they sit inactive for 10 minutes - in that case I can see this error  
making sense since the lockfile is on a different machine.

Note that it's a Horde error.  Imp throws a similar error but ONLY on  
logout:

Aug 29 13:42:30 webcluster3 HORDE[25341]: [imp] Logout for someuser  
[xxx.xxx.xxx.xxx] from {yyy.yyy.yyy.yyy:143} [on line 42 of "/var/www/ 
horde/imp/login.php"]
Aug 29 13:42:30 webcluster3 HORDE[25341]: [imp] Error retrieving  
session data (id = 28b9358a717b995ed145f74972ed2ccc) [on line 83 of "/ 
var/www/horde/lib/Horde/SessionHandler/memcache.php"]

In this case both process IDs are the same.  My guess is that it's a  
race condition somewhere that is asking for the session data a split  
second after the session has been expired.  When Horde throws the  
error, though,  there is no corresponding entry with the same process  
ID.

Now this doesn't appear to be causing any problems for users but it  
IS effectively doubling my log sizes with noise.

So if anyone else is doing it similarly and can tell me if this is   
normal or if there's something else that needs to be tweaked that'd
be great!

Thanks!!