Avoid using root path on admin UI

[v9n]

I want to discuss this change. If you guys are ok, I'm happy for a PR.

Currently, admin UI on host:8083 use absolute asset path. On some strict environment, I cannot open port 8083, I therefore using Nginx and set it up like this: http://host/influxdb_admin -> 127.0.0.1:8083. However, the asset cannot be loaded because they will be set to host:8083/asset. Note that, port 8083 isn't open to public.

I suggest that we removed that and respect that uri and prepend them to asset path propertly to make above setting work.

[mei-rune]

+1

[xrl]

The admin UI will also try making API calls to that root URL. The toml file should have an entry for admin base URL and assets/api calls should built off that.

[v9n]

@xrl API calls make to that URL on port 8086. When I use same technique such as host/influxdb_apu -> 127.0.0.1:8086 the API call work fine. I tried to change asset on config file but it doesn't have any effect. I wil try more now.

[begriffs]

+1

I would like to proxy on a nested root path as well.

guys, do you have a plan to introduce this functionality within the next major release 0.9 version?

Could anyone help to solve authentication issue? It goes to http://IP:8086/cluster_admins rather than http://IP:80/influxdb_web/cluster_admins (we defined /influxdb_web as Nginx custom location)

[mikhailov]

guys, is there any option to use Nginx as a reverse-proxy for InfluxDB web/api access and avoid using CORS? Basically it is important by security and SSL offloading options.

[pauldix]

not currently. We'll see what we can do for 0.9.0. Out of curiosity, why do
you want to use Nginx as a reverse proxy?

On Tue, Jan 20, 2015 at 1:16 PM, Anatoly Mikhailov <notifications@github.com

wrote:

guys, is there any option to use Nginx as a reverse-proxy for InfluxDB
web/api access and avoid using CORS?

—
Reply to this email directly or view it on GitHub
#1236 (comment).

[xrl]

Nginx is great for having uniform access controls (LDAP integration) and also encapsulating deployment complexity. I make a landing page which links to /graphite, /grafana/, /kibana, and now I'd like to have /influxdb. Nginx in turn hides how all those are services are running.

[v9n]

@pauldix I totally agree with @xrl
Also, we cannot open other port than port 80 bcuz a strict firewall and have to use HAProxy/Nginx. HAProxy/Nginx detect the host/domain and route into correct server/backend. It's much more secure than exposing InfluxDB directly IMHO. Dealing with SSL and other magic stuff in Nginx/HAProxy too.

[mikhailov]

yes, @xrl and @kureikain are absolutely right, by some security concern (security audit requires to have 80/443 ports open only for app nodes, for instance), we can keep SSL certificates settings in one place (Nginx configuration), we can load balance between machines to build failover for influxDB web interface and the API. Many many different use cases where reverse-proxy makes sense.

[porjo]

This is a duplicate of https://github.com/influxdb/influxdb-admin/issues/34. I will close that issue in favour of this one.

[mikhailov]

@porjo, will you merge the commit https://github.com/Sean1e/influxdb/commit/fe0967c6ab5d672ec896a3120cd0ef668e4258ff ? It also required to recompile all static assets.