Damon,

No problem. I am glad things are working for you now.

You were also right, we have another customer who started experiencing this problem after upgarde to 3.7.x.

Cheers,
Anton

Oops, yes turns out ProxyPreserveHost does work for us after all.. I had a broken mental model about what was supposed to happen.. Thanks for your help!

Damon.

Hi Damon,

After some discussions here I also believe the proxyName and proxyPort setting will fix the problem.

The only solution that will work for us is a resolution on the TC box

I was more refering to investigating in a test environment to see if it is Apache or the load balancer that is causing the problem.

PS. You may find others with this problem as more people upgrade.

You are right. However in JIRA 3.6.5 we had people who had problems with RSS feeds when accessing JIRA through different IPs. JIRA 3.7 solved that problem, but introduced this one. As I hope that setting proxyName and proxyPort will fix this problem, I believe the current way things are done is the best alternative.

Hi,

Unfortunately there is no way to simplify our network configuration.. Thats a different team and it ain't gonna happen.

The only solution that will work for us is a resolution on the TC box.. From a brief look at the docs you provide the proxyName and proxyPort may work for us.. We'll give it a try and see how it goes..

Damon.
PS. You may find others with this problem as more people upgrade. RSS feeds worked fine for us in 3.6.5 so it could just be that some customers with complex networks like ours are yet to upgrade...

Hi Damon,

I am not a huge expert on load balancers. However, I hope the firewall should not get in the way too much here as it mostly acts as a packet filter. If you try to configure Apache as suggested in our documentation, does it help?

The baseurl is used for all other links in Jira. Why not use it for RSS feeds as well?

The reason is that there is only one baseurl in JIRA. And we have had quite a few customers having problems with RSS feeds as they access JIRA via different IPs. If we use a baseurl for everybody, for some users the URLs do not work. Therefore we are trying to change JIRA such that the baseurl is only used when we are producing content without a HTTP request (e.g. e-mails, as these are produced asynchronously from a background thread).

Would it be possible to simplify the network environment such that we can narrow down what actually causes a problem? That is, whether it is Apache, firewall or load balancer.

It might also be possible to force the correct IP address to be used by using the proxyName and proxyPort parameters as suggested in the JIRA-Apache documentation:
http://www.atlassian.com/software/jira/docs/v3.7.2/apacheintegration.html

I am not 100% sure on this, as I have very little experience with load balancers. But if using the ProxyPreserveHost does not work, then giving proxyName and proxyPort a shot might be a good idea.

May I also ask why the requests to JIRA go through a load balancer? JIRA cannot be clustered and therefore there should be only one instance of JIRA running at any one time.

If you are still having trouble, I believe the correct thing to do is to create a support request in our support system:
http://support.atlassian.com

Thanks,
Anton

Hi Anton,

Thanks for the reply.. Yes, our network config is a little complex.. Jira sits behind first a load balancer, then a hardware firewall, then an Apache reverse proxy then it hits our Tomcat box.. Each step along the way changes the IP address.. Our public IP is so far removed from the actual Tomcat box that there will be no way for us to pass a public IP all the way to the backend I suspect..

The baseurl is used for all other links in Jira. Why not use it for RSS feeds as well?

Regards,
Damon.

Chris,

The IP address should be the public IP address of the machine that hosts JIRA, which was actually used to fetch the RSS feed. Sometimes a machine where JIRA is hosted has multiple IP addresses, and different users access JIRA from different locations via different IP addresses. That is, JIRA is "multihomed". For each user, the link should display the IP address that was used to access JIRA. We have quite a few customers running a "multihomed" JIRA.

May I ask if JIRA is behind Apache HTTP Server or IIS? If so, you will need to configure them correctly to pass through the information about the IP address the user actual used to access JIRA. For more information see the following documentation:
http://www.atlassian.com/software/jira/docs/v3.7.2/apacheintegration.html
(especially the part that talks about the "ProxyPreserveHost" directive).

I will resolve this issue for now. If you are still having troubles, please let us know.

Thanks,
Anton