Forum Discussion
Jacob_W_102765
Nimbostratus
NimbostratusCan't get traffic through port 8080
I hate asking, because I like to solve these on my own, but I need help from someone with more experience than me.
Main Server: main.univ.edu
CNAME for Main Server: alias.univ.edu
We have an application that uses it's own tomcat server. When you go to alias.univ.edu:8080, everything works fine.
Right now, I have a redirect rule that takes 'alias.univ.edu' to 'alias.univ.edu:8080' and this works. I am tasked with getting rid of the ':8080' so that when a user puts in the address 'http://alias.univ.edu', it hits the tomcat app but shows only the alias.univ.edu in the address bar and, of course, everything has to work properly.
I have set up a VS and have NAT-ed an internal IP.
NAT-IP: 123.123.123.246
Origin: 21.21.21.42
Virtual Server:
-Partition: Common
-Destination: 123.123.123.246
-Service Port: 80
-Type: Standard
Default Pool: *Assigned*
No Default or Fallback Persistence Profiles
POOL
Partition: Common
Availability: Available (Enabled)
Configuration: Basic
Health Monitor: http
*Only one pool member enabled. One other pool member set on port 80 appears okay while apache is running on the server.
Pool Members:
Address: 21.21.21.42
Service Port: 8080
Parent Node: 21.21.21.42 *(Green circle)
Availability: Offline (Enabled)-Pool member has been marked down by a monitor *(Red diamond)
Health Monitor: http *(Red diamond)
Configuration settings (Basic) are set to default settings.
Statistics show that there have been no connections and no transfers.
Any help would be appreciated.
3 Replies
- The_BhattmanHi Jacob,
The domains are pointed to single IP address and it's working on 8080, then I am assuming that you have a VIP on 8080 as well?
Bhattman 
hoolioCirrostratus
I think Jacob wants to not include port 8080 on what's presented to clients. So the VIP should be defined on port 80 and the pool member(s) on port 8080.
If the monitor is marking the pool member down, then I'd look at the monitor configuration. I wouldn't suggest using a default HTTP monitor as it doesn't use a receive string. So any response from the server would be considered a success. Also, the send string is an HTTP 0.9 request--which many web servers won't handle.
Can you try setting a send string of:
GET / HTTP/1.1\r\nHost: \r\nConnection: close
and a receive string of "HTTP" to start with?
If you want more tips on troubleshooting monitors, you can check this page:
http://devcentral.f5.com/Wiki/default.aspx/AdvDesignConfig/TroubleshootingLtmMonitors.html
Aaron
Aaron- hoolioGlad that worked for you. Once you confirm the monitor works with a receive string of HTTP, it would be ideal to check for a specific response code like "200 OK". This ensures that the server is responding with a success code--not just any HTTP response.
Also, you don't need to specify an alias port of 8i080 or create a custom monitor for port 8080. The monitoring daemon will poll the pool member IP:port, so you can create a custom HTTP monitor without an alias IP or port and use it on the port 8080 pool members.
Aaron
