> > We write enterprise level software probably far surpassing > what PHP was ever intended for. > > > > However, our GUI is web based (LAMP). > > so the GUI uses php but the rest doesn't? > kind of nullifies the statement aboveif thats the case. Not really. We use DBUS calls to Ruby and C/C++ code. We manipulate networks. TCP/IP. UDP. LDAP. Iptables. Etc. We use RDBMS tricks to transfer data. SOAP. XML. And all sorts of other tactics to work around the limitations of PHP5. But that is all besides the point. > > We have fail over cluster nodes. If a user is logged into > one via a virtual > > IP, the browser sees it as transparent. When a node fails, > it fails over > > fine (again, the browser still sees the same VIP). But the > sess_ file is not > > on the new node -- by design. We purposely don't copy the > /tmp/sess_ files. > > What we want is, since the session is gone, that > $_SESSION['login'] is (in > > theory) missing/false [although it seems that PHP RAM takes > precedence over > > HD now and this didn't used to be the case] that the user should be > > re-prompted to login. > > so the user is prompted to login in again if a cluster node > he happened to > be talking to fails.... whats the point of the transparency > then? I really don't > care if my browsers sees the IP consistently - I'd rather > just stay logged in. Because. That's also irrelevent. And for the record. They DO stay logged in. That's the whole problem I'm trying to get around! For various reasons beyond the scope of this discussion (some related to security), We wish for the user to re-authenticate. d -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
