2

I have a working configuration like this:

+-----+    +-----+    +-----+     +-----+
|  i1 |    |  i2 |    | i3  | ... | in  |    JBoss Cluster
+--+--+    +--+--+    +--+--+     +--+--+
   |          |          |           |
   +----------+-----+----+----...----+
                    |
                +---+---+
                | httpd |
                +-------+
                    |
                   WAN

That is, the httpd balancing the load and providing HA with the JBoss cluster. Now, I need to provide HA for the httpd. If the httpd server goes down, it should be backed with another doint the same job (LB and HA within JBoss cluster).

I'm not sure how session data will be managed, and if simple HA configuration between the 2 httpd instances will work ok.

Does anyone face a situation like this?

Sebastian
  • 121
  • 2
  • How are you planning on switching between `httpd` instances? – Xiong Chiamiov Dec 12 '13 at 23:11
  • @XiongChiamiov, if I understood you correctly: that depends wether or not the `httpd` server is down. I want to provide high availability for `httpd`, but to keep the cluster configuration working (implying session data and so on). For example: one `httpd`instance could be master. Another one, a slave. If master server goes down, the slave should receive and forward all request to the JBoss cluster. – Sebastian Dec 13 '13 at 14:47
  • Right. So, do you have a plan for how you're going to monitor `httpd` and do failover, or is that part of what you want to know? Sidenote: sessions are usually managed in your application, not within Apache. – Xiong Chiamiov Dec 21 '13 at 01:27
  • @XiongChiamiov heartbeat seems a good approach. But I don't know if a particular approach should be taken considering that `JBoss` servers are behind httpd, or, it will be transparently the HA between the httpd instances no matter what we are serving. Am I clear? – Sebastian Dec 23 '13 at 18:36

0 Answers0