Confluence and JIRA got desynchronized after modifying IP address range

[guillempons]

Hello,

We are a small team and we are using Jira and Confluence (latest versions, updated a month ago). We are using Jira's feature "JIRA user server", so that all the users from Jira can also access to Confluence with the same user.

Last day we modified the IP range of the whole office (from 192.168.1.X/24 to 10.12.0.X/24) and Jira works perfectly fine but confluence not. The domain names have not been modified.

First of all I could not login as administrator (it should not happen because although users are synchronized, there's always a local admin or at least I though that). I followed this guide and modified successfully the password, so now I can enter with the local administrator account.

Now the problem is the synchronization with the other users. So, what I have checked so far:

1. Under Jira > Applications > Application Links I have linked the confluence with Connected status.
2. Under Jira > Applications > Jira user server I have linked the confluence with the correct IP address.
3. I have temporary disabled the Whitelist for Confluence and Jira so as to make sure that was not the problem.
4. Under Confluence > Application Links I have linked the jira with Connected status

Other details:

1. We are using Centos7.
2. Jira and Confluence instances are on different machines
3. We are using MariaDB as database.
4. The other users are not inside any of the confluence-* groups, but cannot add them to the groups because an error pops up (I suppose it is because confluence cannot sync with jira).
5. We are not using HTTPS.

Then, when I test the connection from confluence to jira, this error pops up:

This is the first time I see this "Crowd Server", to be honest. We have not bought this sever nor configured it in the past. Anyway, if I try to add a new Directory (Confluence > User directories) and select Atlassian Jira the error is the same.

Thanks in advance ;)

[Craig Castle-Mead · Accepted solution]

Hey Guillem,

While Crowd is a standalone product, each of the main Atlassian applications (jira/conf/bb etc) run "embedded Crowd" for user management which is why you're seeing it referenced.

Trying to authenticate to a Crowd application with a bad username and password returns a HTTP 401, but get the user/pass right but from a different source IP and you get a 403 (which is what you're getting).

You did mention in your3rd point that you'd "disabled the Whitelist for Confluence and Jira" but it really feels like this is the most likely issue, so want to confirm what you did.

1. Go to https://jira.mycompany.com/secure/admin/ConfigureCrowdServer.jspa
2. Edit your existing application (likely Confluence)
3. Add the following to the "IP addresses" textbox:
   1. *.*.*.* (allow everything) - see if you can the login from Confluence
   2. If that works, change the *.*.*.* to 10.12.0.0/24 and check again
   3. Confirm the Confluence servers IP and change to it (eg: 10.12.0.10 ) to make it more secure

If you can, check the atlassian-jira.log for a line containing the string "Client with address" and it should tell you what the Confluence server is reporting to Jira as. 

CCM

[guillempons]

Thanks, it's working. I love you!

Instead of 10.12.0.X/24 I wrote 10.12.0.X, so I suppose Confluence acted normally and used the /8 range instead of the /24. Such a stupid error, sorry for wasting your time Craig.