Hi Folks,
We are in the final stages of implementing a secondary level Layer 7 switch between our reverse proxy layer and our dual server app layer and we need to know if Matrix is a Statefull or Stateless application.
Why do we need to know: So we can configure the switch to either be session sticky (that is a session only ever resides with the one server) or if a session can be passed around between the 2 app servers.
The answer will revolve around where session information is stored…
Look forward to the response/s :lol:
-=Anthony=-
Matrix is sticky unless you have shared storage between your web servers. The session information is stored in MATRIX_ROOT/cache (as sess_* files).
Can I ask how your two application servers are currently configured?
[quote]Matrix is sticky unless you have shared storage between your web servers. The session information is stored in MATRIX_ROOT/cache (as sess_* files).
Can I ask how your two application servers are currently configured?
[right][post=“13580”]<{POST_SNAPBACK}>[/post][/right][/quote]
Hi Avi,
Our app servers are Sun T2000 servers. These T2000 have internal disk for host OS and access to a “Shared” SAN based FileSystem for the application data. The application data disk is presented to the app servers via a NFS mount points The San disk is made up of multiple parity groups in the San that are then ZFS stripped to gain max efficiency of the disk space…
Hope this answers your question 
With regards to the statefullness of Matrix, I would assume then that for us Matrix shall be a stateless application. This means that the Switch can manage traffic across the 2 servers with worrying about sticky connections…
A
We've had some issues with T2000 boxes and Apache segfaulting. You're not experiencing any problems like this, are you?
[quote]We’ve had some issues with T2000 boxes and Apache segfaulting. You’re not experiencing any problems like this, are you?
[right][post=“13586”]<{POST_SNAPBACK}>[/post][/right][/quote]
Hi Avi,
Currently no problems at all with seg faults. We have compiled apache2 from source, and optimised for T2000 environment. Also we are using zones/containers on the T2000 to manage the application and it’s resource grabbing…
For all intents and purposes though most of our traffic is managed by the 4 Reverse Proxy servers (sun v240r’s) that are running Squid 2.6 and again compiled form source and optimnised for the T2000 environment…
So far we are very happy with the environment and the speed of the system. I have yet to sit down and do formal tuning of the system, this shall occur in the New Year and I shall post our results here.
:)Anthony
Sounds awesome.