[
https://issues.apache.org/jira/browse/COUCHDB-2343?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14329004#comment-14329004
]
ASF subversion and git services commented on COUCHDB-2343:
----------------------------------------------------------
Commit 2a583cb0dfcd446ae259b272acd58068079c9b52 in couchdb-chttpd's branch
refs/heads/master from [~robertkowalski]
[ https://git-wip-us.apache.org/repos/asf?p=couchdb-chttpd.git;h=2a583cb ]
Remove _config route on cluster
In order to avoid users shooting themselves in the foot by using
`/_config/` on a clustered CouchDB with a loadbalancer in front,
we remove it on `15984` - it will be available for single-node-
mode on the backdoor port (`15986`) or for users that are feeling
lucky which want to fire curl requests to every node.
It also allows Fauxton to detect if it is running on a the backdoor
port. Fauxton will - if it gets a 200 instead of a 404 - show the
config-section to the user.
COUCHDB-2601 COUCHDB-2390 COUCHDB-2343
> /_config/admins/username fails on master
> ----------------------------------------
>
> Key: COUCHDB-2343
> URL: https://issues.apache.org/jira/browse/COUCHDB-2343
> Project: CouchDB
> Issue Type: Bug
> Security Level: public(Regular issues)
> Components: HTTP Interface
> Affects Versions: 2.0.0
> Reporter: Joan Touzet
> Priority: Blocker
> Labels: auth
> Fix For: 2.0.0
>
>
> In a multi-node setup, calling _config/admins/username to create an admin
> user fails to correctly configure a cluster with a new administrator. This
> fails for two reasons:
> 1) The call is only processed on a single node, and the admin entry is not
> replicated
> 2) Even if the call is repeated on all nodes manually, the hashes will be
> different on each node, which will cause cookie failure when attempting to
> authenticate via other machines.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
