I'm afraid this issue is back again for Jira 7.4.1 and I'm running with Confluence 6.3.1. When my application links are setup with Oauth,(Impersonate) and I select 'Linked Pages' on the Sprint backlog board, it shows link: "Page failed to load" but if I select the link it takes me to the correct Confluence page and with no authentication prompts. If I switch the Application Links to just Oauth and I select the 'Linked Pages' , a dialog box pops up asking me to authenticate and then asking to Allow/Deny the linkage. Once I do that, the "Linked Pages" properly shows the title of the linked Confluence page.
As a workaround, it recommends to disable JIT optimization. I will try it but this reflects poorly on Atlassian's regression testing that this issue keeps reoccurring.
Sorry to hear that you are seeing this problem once more on a newer version of JIRA. I would like to investigate this problem further to better understand if this is the same bug that is causing this issue, or if there is something else that might be causing this behavior.
In order to do so, I create a support request on your behalf and requested some log files so that we can dig into this more.
In the previous support case we found there was a problem in regards to that user accounts permissions in Confluence. The account in question was not able to view the space that page was located in over in Confluence. Hence the link also cannot be loaded within Jira when that happens.
Of course there could be other problems to cause this behavior as well. But I would recommend checking first to confirm that your user account has access to both applications to see the linked data.
In the previous case, the Application link was setup to use OAuth (with impersonation). This has an additional implication. Namely, that the user accounts in both Jira and Confluence are exactly the same (same username and same password in both applications). This is how Cloud app links are managed because we know the userbase is always the same within a single Atlassian Cloud site. But in the Server/Data Center platform of our products, we can't make that assumption.
If you have a different username for either application and are using OAuth with impersonation for that applink, then that is one way in which this kind of integration can fail. However I am not certain that is what is exactly happening in your case.
If you have the ability to, I would suggest creating a support case with us to more closely investigate your environment. If you are not able to do that within our Support offerings, then I would be interested to take a closer look at the logging of Jira when loading that page. I suspect it is trying to make a call to the Confluence application when this page is loaded. Should it be failing I would look at the $JIRAHOME/log/atlassian-jira.log file at that time to see if perhaps you can find any additional clues about if there might have been a WARN or ERROR level message that could point us in a more concise direction for troubleshooting this problem.