Administrative/direct access to Bitbucket node behind loadbalancer

[Deleted user]

Hi,

setup:
loadbalancer: git.company.com:443
node1: bitbucket01.internal.company.com:7990
node2: bitbucket02.internal.company.com:7990

We just upgraded from Bitbucket DC 4.14 to 7.3.1 including the migration from server.xml to bitbucket.properties.

With the old version we were able to directly access the nodes with http://bitbucket01.internal.company.com:7990. After the upgrade we immediately get redirected to "https://git.company.com" so its not possible to see/troubleshoot the node startup and to access a node directly.

Is it possible to get the old behavior back?

bitbucket.properties:

[...]
server.secure=true
server.scheme=https
server.proxy-port=443
server.proxy-name=git.company.com
[...]

curl:

user@f00:~$ curl -I http://bitbucket01.internal.company.com:7990/
HTTP/1.1 302 
[...]
Cache-Control: no-store
Location: https://git.company.com/repos?visibility=public
Content-Language: en-US
Transfer-Encoding: chunked
Date: Wed, 08 Jul 2020 13:03:19 GMT

[Sujay]

Hi,

You can enable an additional connector with a port other than 7990 to achieve this. Follow the instructions in https://confluence.atlassian.com/bitbucketserverkb/how-do-i-bypass-a-proxy-for-bitbucket-server-826878770.html to create a connector with port 7995 and you should be able to connect as follows:

loadbalancer: git.company.com:443
node1: bitbucket01.internal.company.com:7995
node2: bitbucket02.internal.company.com:7995

7990 will hit a redirect to 443 and reach git.company.com, so you will not be able to use it.

Feel free to contact us in case of any questions.

Cheers,

Sujay Raj

Atlassian Support 

[Deleted user]

thx, that sounds like a solution.

Would be nice if https://confluence.atlassian.com/enterprise/load-balancer-configuration-options-935383760.html has a link to that page in the "Administrative access" section.

Thanks and Regards

[Sujay]

Hi again,

This would not be a recommended practice because if you have direct access to the nodes, then there is a risk of one node being flooded with requests by direct access while the others are idle. Bypassing the proxy and accessing a node directly is used mostly for troubleshooting issues and which is why it's part of a KB.

I think that's why you don't see it mentioned in the documentation you mentioned. Hope that helps.

Feel free to contact us in case of any questions.

Cheers,

Sujay Raj

Atlassian Support