according to a #46 need to add such functionality.

We need to add mechanism for discovery linked containers
this is a common use case for developers , regular users etc..
I will add more info here later

I believe there must be "+=" here

Keep optional plaintext password as wel

Hi all,
I have one question if you don't mind,
I see that the API supports TLS connection, but what about the TCP reverse proxy? Will you support TLS in the future?

It would be nice to let backends warm up for configured time after discovery have added them.

server:

• active connections
• total connections
• traffic rx/tx (total)
• current traffic rx/tx (gb,mb,kb,b,/s)
• backends
  • status : live | down
  • inbound traffic (total/avg(1 sec))
  • outbound traffic (total/avg(1 sec))
  • active connections
  • total connections

lb configured with :
client_idle_timeout = "1h"
backend_idle_timeout = "1h"

but each hour connections dropped (long connections with continuous static bandwidth data flow)

version : 0.3.0 release

Due to security reasons add optional ability to use TLS and client authentication while querying Docker API, providing additional parameters in docker discovery config.

Proposed format:

[servers.<name>.discovery]
# ...
kind = 'docker'
# ...
docker_tls_enabled = false # false by default for backward compatibility

# required if docker_tls_enabled = true
docker_tls_cert_path = '/path/to/cert.pem'
docker_tls_key_path = '/path/to/key.pem'
docker_tls_cacert_path = '/path/to/cacert.pem'

More info: https://docs.docker.com/engine/security/https/

For future releases we can add mesos/marathon support via their rest api.

https://mesosphere.github.io/marathon/docs/generated/api.html#

api enteryponint - ip/hostname:port
api service enterypoint - /v2/apps/[app_id]/tasks

Response:

Body

Type: application/json

Example:

{
  "tasks": [
    {
      "appId": "/minecraft/survival-world",
      "host": "srv7.hw.ca1.mesosphere.com",
      "id": "minecraft_survival-world.564bd685-4c30-11e5-98c1-be5b2935a987",
      "ports": [
        31756
      ],
      "slaveId": null,
      "stagedAt": "2015-08-26T20:23:39.463Z",
      "startedAt": "2015-08-26T20:23:44.678Z",
      "version": "2015-04-17T04:00:14.171Z"
    }
  ]
}

Type: text/plain

minecraft_survival-world    10013   srv7.hw.ca1.mesosphere.com:31756

to share/healthcheck

• mongo
• postgresql
• mysql
• redis
• http/https
• anything else

Basic functionality for gobetween frontend :

1. password auth
2. servers list
3. Stats sheet/graph  for each server 
3.1 for server : 
     incoming connections quantity,
     active connections
     traffic (rx/tx) 
3.2 for each backend node:
      check status (ok,failed)
      connection quantity    
      active connections
       traffic (rx/tx)
       weight
       priority
4. gobetween configuration constructor
     basic functionality
     add/delete servers
     add/delete backend nodes for static list
     change discovery type for each server/edit backend variables
     change check type /check variables 

list is unfinished....

steps to reproduce
1)start gobetween without any servers configured
2) add server via api
3) add same server second dime
4) possible need to continue step 3)
5) check ip:apiport/dump and ip:apiport/servers//stats
some of them or both links stops to respond

As user i would like to have an option to return private container IP and service Port pairs during discovery for cases where Docker Networking is used and load balancer can reach services directly (without port mapping).

When amount of SRV records in service is 6 or more discovery fails with error:

2016-07-19 18:21:39 [INFO ] (srvFetch): Fetching consul.service.consul:8600 nginx.service.consul.
2016-07-19 18:21:39 [ERROR] (discovery): srv error dns: buffer size too small retrying in 2s
2016-07-19 18:21:39 [INFO ] (discovery): Applying failpolicy keeplast

After service has been scaled down, problem went away

netstat -anp |grep 9200 |wc -l

178

{
    "active_connections": 6,
    "rx_total": 1562227305,
    "tx_total": 1724183714,
    "rx_second": 21308,
    "tx_second": 55111,
    "backends": [
        {
            "host": "10.0.0.53",
            "port": "9200",
            "priority": 1,
            "weight": 1,
            "stats": {
                "live": true,
                "total_connections": 9692,
                "active_connections": 1,
                "rx": 234993851,
                "tx": 133001288,
                "rx_second": 187,
                "tx_second": 318
            }
        },
        {
            "host": "10.0.0.54",
            "port": "9200",
            "priority": 1,
            "weight": 1,
            "stats": {
                "live": true,
                "total_connections": 9839,
                "active_connections": 1,
                "rx": 216484944,
                "tx": 131607080,
                "rx_second": 0,
                "tx_second": 491
            }
        },
        {
            "host": "10.0.0.55",
            "port": "9200",
            "priority": 1,
            "weight": 1,
            "stats": {
                "live": true,
                "total_connections": 9629,
                "active_connections": 1,
                "rx": 166526541,
                "tx": 2506771,
                "rx_second": 39,
                "tx_second": 32
            }
        },
        {
            "host": "10.0.0.51",
            "port": "9200",
            "priority": 1,
            "weight": 1,
            "stats": {
                "live": true,
                "total_connections": 9849,
                "active_connections": 2,
                "rx": 776211040,
                "tx": 1453758841,
                "rx_second": 34081,
                "tx_second": 80542
            }
        },
        {
            "host": "10.0.0.52",
            "port": "9200",
            "priority": 1,
            "weight": 1,
            "stats": {
                "live": true,
                "total_connections": 9508,
                "active_connections": 0,
                "rx": 168079725,
                "tx": 3470986,
                "rx_second": 39,
                "tx_second": 32
            }
        }
    ]
}

# dumped on 2016-07-13 04:55:44.270808288 +0000 UTC

[logging]
  level = "info"
  output = "/var/log/gobetween.log"

[api]
  enabled = true
  bind = "10.0.0.8:8888"

[defaults]
  max_connections = 0
  client_idle_timeout = "0"
  backend_idle_timeout = "0"
  backend_connection_timeout = "0"

[servers]
  [servers.elasticsearch]
    max_connections = 10000
    client_idle_timeout = "1m"
    backend_idle_timeout = "1m"
    backend_connection_timeout = "3m"
    bind = "10.0.0.8:9200"
    protocol = "tcp"
    balance = "weight"
    [servers.elasticsearch.discovery]
      kind = "static"
      failpolicy = "keeplast"
      interval = "0"
      timeout = "0"
      static_list = ["10.0.0.51:9200", "10.0.0.52:9200", "10.0.0.53:9200", "10.0.0.54:9200", "10.0.0.55:9200"]
    [servers.elasticsearch.healthcheck]
      kind = "ping"
      interval = "10s"
      passes = 3
      fails = 3
      timeout = "1s"

last logs

2016-07-12 22:00:16 [INFO ] (proxy): read tcp 10.0.0.8:38724->10.0.0.53:9200: use of closed network connection
2016-07-12 22:00:18 [INFO ] (proxy): read tcp 10.0.0.8:38281->10.0.0.52:9200: use of closed network connection
2016-07-12 22:00:55 [INFO ] (proxy): read tcp 10.0.0.8:57795->10.0.0.55:9200: i/o timeout
2016-07-12 22:00:55 [INFO ] (proxy): read tcp 10.0.0.8:9200->10.0.0.60:45876: use of closed network connection
2016-07-12 22:01:13 [INFO ] (proxy): read tcp 10.0.0.8:59927->10.0.0.53:9200: i/o timeout
2016-07-12 22:01:13 [INFO ] (proxy): read tcp 10.0.0.8:9200->10.0.0.60:45347: use of closed network connection
2016-07-12 22:01:14 [INFO ] (proxy): read tcp 10.0.0.8:35520->10.0.0.54:9200: i/o timeout
2016-07-12 22:01:14 [INFO ] (proxy): read tcp 10.0.0.8:9200->10.0.0.60:45348: use of closed network connection
2016-07-12 22:01:18 [INFO ] (proxy): read tcp 10.0.0.8:9200->10.0.0.60:45344: i/o timeout
2016-07-12 22:01:19 [INFO ] (proxy): read tcp 10.0.0.8:9200->10.0.0.60:45345: i/o timeout

Instead of swarm that respond with :

ec9ec5753ee2        test1/company:test   "/bin/bash -c 'start   46 hours ago        Up 46 hours         5.5.5.5:39599->9999/tcp, 5.5.5.5:50300->9000/udp,  5.5.5.5:39600->9000/tcp         n1.node.test.com/12121212

docker node

ec9ec5753ee2        test1/company:test   "/bin/bash -c 'start   46 hours ago        Up 46 hours         0.0.0.0:39599->9999/tcp, 0.0.0.0:50300->9000/udp, 0.0.0.0:39600->9000/tcp         n1.node.test.com/12121212

So, by default in case of single remote docker host discovery we can`t determine docker containers IP.
It can be fixed in Linux by running :

docker run -d -p=192.168.123.138:80:80 tutum/hello-world

but under windows we have :

PS C:\Docker> docker -H tcp://127.0.0.1:2375 run -l proxied=true -d -p 10.0.0.4:8083:80 helloworld
edd0a1ea04851d5d0c817efb2fd3ea0ee6e3f31f89b3fb53a220ab07feead00b
C:\Windows\system32\docker.exe: Error response from daemon: failed to create endpoint hungry_ptolemy on network nat: Windows does not support host IP addresses in NAT settings.

 PS C:\Docker> docker -H tcp://127.0.0.1:2375 run -l proxied=true -d --ip 10.0.0.4 -p 8083:80 helloworld
f4069038a829d617718bc80e4b33fb3a10468fc155b1742dba5319653703650b
C:\Windows\system32\docker.exe: Error response from daemon: User specified IP address is supported on user defined networks only.

it needs to eliminate this case for windows compatibility and allow to use linux docker also with wildcard binding

for this we need check does discovery return 0.0.0.0:port and change port with variable :
docker_binded_host=IP/hostname

need to add such functionality via api without deleting/recreating server:

1. disable backend server ip:port (continuously). This should disable all traffic to this node and exclude it from active backend pool (same as for faulty healthcheck)
2. Enable backend server ip:port (continuously). This should enable(rebalance according to a policy) traffic to this node and exclude it from active backend pool (same as for healthy healthcheck)
3. if this server disappear from backend pool - delete it`s "disable" flag.

Hi ALL

For build gobeetwen at the clear system add this lines into Makefile

go get github.com/BurntSushi/toml
go get github.com/Sirupsen/logrus
go get github.com/elgs/gojq
go get github.com/fsouza/go-dockerclient
go get github.com/gin-gonic/gin
go get github.com/hashicorp/consul/api
go get github.com/miekg/dns
go get github.com/spf13/cobra

Hello,
Im trying gobetween on my dev machine - Im using docker-machine & boot2docker (1.12.3)

This is my simple docker compose:

version: '2'

networks:
  backend:

services:
  app:
    image: php:7.0-alpine
    expose:
      - 8080
    volumes:
      - .:/data
    command: php -S 0.0.0.0:8080 /data/index.php
    networks:
      - backend
    labels:
      - "scale.app=true"

  gobetween:
    image: yyyar/gobetween
    depends_on:
      - app
    ports:
      - "80:80"
    volumes:
      - "./gobetween/conf:/etc/gobetween/conf/:rw"
      - "/var/run/docker.sock:/var/run/docker.sock"
    networks:
      - backend

Gobetween conf looks like this:

[logging]
level = "debug"

[servers.app]
bind = "0.0.0.0:80"
protocol = "tcp"
balance = "roundrobin"

  [servers.app.discovery]
    interval = "10s"
    timeout = "2s"
    kind = "docker"
    docker_endpoint = "unix://var/run/docker.sock"  # Docker / Swarm API
    docker_container_label = "scale.app=true"  # label to filter containers
    docker_container_private_port = 8080   # gobetween will take public container port for this private port
    docker_container_host_env_var = "HOSTNAME"

Gobetween is able to discover new backends, when i try to scale up the app, but proxying is not working - see log output:

app_1        | PHP 7.0.13 Development Server started at Fri Nov 25 20:08:30 2016
gobetween_1  | 2016-11-25 20:08:31 [INFO ] (manager): Initializing...
gobetween_1  | 2016-11-25 20:08:31 [INFO ] (server): Creating 'app': 0.0.0.0:80 roundrobin docker none
gobetween_1  | 2016-11-25 20:08:31 [INFO ] (scheduler): Starting scheduler
gobetween_1  | 2016-11-25 20:08:31 [INFO ] (manager): Initialized
gobetween_1  | 2016-11-25 20:08:31 [INFO ] (api): API disabled
gobetween_1  | 2016-11-25 20:08:31 [INFO ] (dockerFetch): Fetching unix://var/run/docker.sock scale.app=true 8080
gobetween_1  | 2016-11-25 20:08:40 [DEBUG] (server.handle): Accepted 10.211.55.2:52649 -> [::]:80
gobetween_1  | 2016-11-25 20:08:40 [DEBUG] (server.handle): Accepted 10.211.55.2:52648 -> [::]:80
gobetween_1  | 2016-11-25 20:08:40 [ERROR] (server.handle): dial tcp :0: getsockopt: connection refused
gobetween_1  | 2016-11-25 20:08:40 [ERROR] (server.handle): dial tcp :0: getsockopt: connection refused

Am I missing something?

To support large DNS lookup responses with >3 records.
Make it backward compatible.

Hi,
Would it be possible to add a feature where gobetween can register itself in the Consul it would
use for service discovery.
This would helpful when the IP of the Load Balancer may change across reboots.

As of now I am making do with a hackish workaround ...
A consul agent on the load balancer machine. and a cron job which get the IP of eth0 and recreates
a service definition in '/etc/consul/'.

Regards,
Shantanu

We store backends as a map in scheduler and when passing it to balancer as a list, due to maps nature, they arrive in random order.
It also affects other balancing types.

exec discovery script :

    @ECHO OFF
    echo webseerver1.loc:80 weight=1
    echo(
    echo webserver2.loc:80 weight=1

discovery config section :

    [servers.sample.discovery]  
    kind = "exec"
    exec_command = ['D:\scripts\exec_discovery.bat'] 
    exec_command = ['node', 'D:\scripts\test.js'] 

Gobetween can discovery only last server from discovery script because of a newline character difference between windows and linux.

Need to make discovery regexp for exec discovery configurable.

Add build-in REST API server functionality

• list / get / create / remove servers
• save current config / restore config from config path selected during startup

2016-07-12 13:28:05 DEBUG: Accepted 10.0.0.50:57617 -> 10.0.0.8:9200
2016-07-12 13:28:05 DEBUG: Begin 10.0.0.50:57617 -> 10.0.0.8:9200 -> 10.0.0.55:9200
2016-07-12 13:28:05 INFO : read tcp 10.0.0.8:49356->10.0.0.55:9200: use of closed network connection
2016-07-12 13:28:05 DEBUG: End 10.0.0.50:57617 -> 10.0.0.8:9200 -> 10.0.0.55:9200
2016-07-12 13:28:05 DEBUG: Accepted 10.0.0.50:57618 -> 10.0.0.8:9200
2016-07-12 13:28:05 DEBUG: Begin 10.0.0.50:57618 -> 10.0.0.8:9200 -> 10.0.0.53:9200
2016-07-12 13:28:05 INFO : read tcp 10.0.0.8:58063->10.0.0.53:9200: use of closed network connection
2016-07-12 13:28:05 DEBUG: End 10.0.0.50:57618 -> 10.0.0.8:9200 -> 10.0.0.53:9200
2016-07-12 13:28:07 DEBUG: Accepted 10.0.0.50:57619 -> 10.0.0.8:9200
2016-07-12 13:28:07 DEBUG: Begin 10.0.0.50:57619 -> 10.0.0.8:9200 -> 10.0.0.51:9200
2016-07-12 13:28:07 INFO : read tcp 10.0.0.8:51749->10.0.0.51:9200: use of closed network connection
2016-07-12 13:28:07 DEBUG: End 10.0.0.50:57619 -> 10.0.0.8:9200 -> 10.0.0.51:9200
2016-07-12 13:28:07 DEBUG: Accepted 10.0.0.50:57620 -> 10.0.0.8:9200
2016-07-12 13:28:07 DEBUG: Begin 10.0.0.50:57620 -> 10.0.0.8:9200 -> 10.0.0.54:9200
2016-07-12 13:28:07 INFO : read tcp 10.0.0.8:33657->10.0.0.54:9200: use of closed network connection
2016-07-12 13:28:07 DEBUG: End 10.0.0.50:57620 -> 10.0.0.8:9200 -> 10.0.0.54:9200
2016-07-12 13:28:07 DEBUG: Accepted 10.0.0.50:57621 -> 10.0.0.8:9200
2016-07-12 13:28:07 DEBUG: Begin 10.0.0.50:57621 -> 10.0.0.8:9200 -> 10.0.0.55:9200
2016-07-12 13:28:07 INFO : read tcp 10.0.0.8:49360->10.0.0.55:9200: use of closed network connection
2016-07-12 13:28:07 DEBUG: End 10.0.0.50:57621 -> 10.0.0.8:9200 -> 10.0.0.55:9200
2016-07-12 13:28:07 DEBUG: Accepted 10.0.0.50:57622 -> 10.0.0.8:9200
2016-07-12 13:28:07 DEBUG: Begin 10.0.0.50:57622 -> 10.0.0.8:9200 -> 10.0.0.53:9200
2016-07-12 13:28:07 INFO : read tcp 10.0.0.8:58067->10.0.0.53:9200: use of closed network connection
2016-07-12 13:28:07 DEBUG: End 10.0.0.50:57622 -> 10.0.0.8:9200 -> 10.0.0.53:9200

seems like it is related to :

http://stackoverflow.com/questions/31385487/golang-http-having-use-of-closed-network-connection-error

./gobetween
2016-07-25 09:05:32 INFO : gobetween v0.2.0
2016-07-25 09:05:32 INFO : Using config file ./gobetween.toml
2016-07-25 09:05:32 INFO : Initializing...
panic: runtime error: invalid memory address or nil pointer dereference
[signal 0xb code=0x1 addr=0x0 pc=0x464d22]

goroutine 6 [running]:
panic(0xa3ca00, 0xc82000a100)
/usr/lib/go/src/runtime/panic.go:464 +0x3e6
_/home/yyyar/workspace/gobetween/src/manager.prepareConfig(0xc8201ce184, 0x6, 0xc8201cd530, 0xc8201cd540, 0xc8201cdb90, 0xc8201cdb70, 0xc8201cc7c0, 0xc, 0xc8201cc838, 0x3, ...)
/home/yyyar/workspace/gobetween/src/manager/manager.go:269 +0x8b2
_/home/yyyar/workspace/gobetween/src/manager.Create(0xc8201ce184, 0x6, 0xc8201cd530, 0xc8201cd540, 0xc8201cdb90, 0xc8201cdb70, 0xc8201cc7c0, 0xc, 0xc8201cc838, 0x3, ...)
/home/yyyar/workspace/gobetween/src/manager/manager.go:116 +0xd7
_/home/yyyar/workspace/gobetween/src/manager.Initialize(0xc8201cc258, 0x5, 0xc8201cc320, 0x6, 0x1, 0xc8201cc4c0, 0x5, 0xc8201d2160, 0x0, 0x0, ...)
/home/yyyar/workspace/gobetween/src/manager/manager.go:46 +0x27b
created by main.main
/home/yyyar/workspace/gobetween/src/main.go:83 +0x4f3

I downloaded the binary package gobetween_0.2.0_linux_amd64.tar

Distributor ID: Ubuntu
Description: Ubuntu 14.04.4 LTS
Release: 14.04
Codename: trusty

uname -a
Linux 3.19.0-61-generic #69~14.04.1-Ubuntu SMP Thu Jun 9 09:09:13 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

For docker discovery need to add multi-endpoint functionality .

something like :

docker_endpoint  = [
    "http://localhost:2375",
    "http://1.2.3.4:2375"
]

Interval and timeout variable values should be applied to each endpoint in pool, not to whole servers pool.

In this case discovery will be asynchronous but it is only one possible deal to implement docker multi-endpoint(multi docker host) discovery. I think it will be okay for most cases.

Provide config option that makes possible to disable health checks.
This makes sense for environments where heath checking is embedded or performed in service discovery tool (e.g. Consul)

And if timeout = 0

2016-07-18 16:10:49 INFO : Responce from exec [/asdda arg1 arg2] is timed out. Killing process...
panic: runtime error: invalid memory address or nil pointer dereference

Hi, Is there any way I can achieve HA for the load balancer?
I would like to have two nodes (doesn't matter AA, or AP) and when a node goes down, the other keeps redirecting traffic.

Thanks.

Add possibility to specify ips / networks to allow / deny access client connections

Need to add consul discovery service integration.

       consul service discovery api calls :
       https://www.consul.io/docs/agent/http/catalog.html#catalog_services

api enteryponint - ip/hostname:port
api service enterypoint -  /v1/catalog/service/[servicename]

[{
    "Node": "ip-172-30-2-11",
    "Address": "172.30.2.11",
    "ServiceID": "ip-172-30-2-11:high_panini:8443",
    "ServiceName": "service1",
    "ServiceTags": ["tag1","tag2"],
    "ServiceAddress": "1.1.1.1",
    "ServicePort": 24338,
    "ServiceEnableTagOverride": false,
    "CreateIndex": 5550785,
    "ModifyIndex": 20803436
}, {
    "Node": "ip-172-30-2-12",
    "Address": "172.30.2.12",
    "ServiceID": "ip-172-30-2-12:insane_kirch:8443",
    "ServiceName": "service1",
    "ServiceTags": ["tag1","tag2"],
    "ServiceAddress": "2.2.2.2",
    "ServicePort": 24138,
    "ServiceEnableTagOverride": false,
    "CreateIndex": 5550914,
    "ModifyIndex": 20803449
}, {
    "Node": "ip-172-30-2-13",
    "Address": "172.30.2.13",
    "ServiceID": "ip-172-30-2-13:happy_shockley:8443",
    "ServiceName": "service1",
    "ServiceTags": ["tag1","tag2"],
    "ServiceAddress": "3.3.3.3",
    "ServicePort": 24338,
    "ServiceEnableTagOverride": false,
    "CreateIndex": 5550996,
    "ModifyIndex": 20803432
}]

Need to parse json output from consul and get pool of :
"ServiceAddress": "ip/hostname"
"ServicePort": port
from there.

Configuration options:

consul_endpoint=”some_host_name_or_ip:port”
consul_service_name=”service_name”
consul_service_tag=”tag” #optional
consul_api_service_endpoint=/v1/catalog/service/[servicename] #optional
consul_api_service_tls_enabled = false # (optional) enable client tls auth
consul_api_service_tls_cert_path = '/path/to/cert.pem'
consul_api_service_tls_key_path = '/path/to/key.pem'
consul_api_service_tls_cacert_path = '/path/to/cacert.pem'

in each healthcheck timeout should be defined and non-zero to prevent errors like this:

2016-07-14 20:29:08 INFO : Responce from exec [/home/ubuntu/lb/exec_healthcheck.sh 54.229.38.77 23756] is timed out. Killing process...
2016-07-14 20:29:08 INFO : Responce from exec [/home/ubuntu/lb/exec_healthcheck.sh 54.229.214.152 22399] is timed out. Killing process...
2016-07-14 20:29:08 INFO : Responce from exec [/home/ubuntu/lb/exec_healthcheck.sh 54.229.214.152 43999] is timed out. Killing process...
2016-07-14 20:29:08 INFO : Responce from exec [/home/ubuntu/lb/exec_healthcheck.sh 54.229.214.152 43989] is timed out. Killing process...
2016-07-14 20:29:08 INFO : Responce from exec [/home/ubuntu/lb/exec_healthcheck.sh 54.229.38.77 32399] is timed out. Killing process...

Hi,
Is there a plan in the near future to implement SNI based routing?

Regards,
Shantanu

BTW:
Awesome software which I discovered only by accident!!!
The Consul discovery backend is truly great!!!
👍 👍

May be useful for REST API consumers. To store some metadata attached.

Need to switch from centos7 base image to https://hub.docker.com/_/alpine/
Image will be much smaller (5 mb base image)

A TLS/SSL termination means handling incoming TLS connections, decrypting the TLS and passing on the unencrypted data to the backends by TCP.

protocol = "tls"

#  [servers.default.tls]             # (required) if protocol == "tls"
#  cert_path = "/path/to/file.crt"   # (required) path to crt file
#  key_path = "/path/to/file.key"    # (required) path to key file
#  min_version = "tls1"              # (optional) "ssl3" | "tls1" | "tls1.1" | "tls1.2" - minimum allowed tls version
#  max_version = "tls1.2"            # (optional) maximum allowed tls version
#  ciphers = []                      # (optional) list of supported ciphers. Empty means all supported. For a list see https://golang.org/pkg/crypto/tls/#pkg-constants
#  prefer_server_ciphers = false     # (optional) if true server selects server's most preferred cipher
#  session_tickets = true            # (optional) if true enables session tickets

client_idle_timeout and backend_idle_timeout parameters does not work as expected. Looks like they are mixed up.

Implement "leastbandwidth" balancer strategy.

Using Docker Discovery Backend, Is there any way to consume different services running same internal container port and different label for example, accessing from same server binded port?.

                    /--containerA:80 LABEL_A
CLIENT ---> HOST:3000
                    \--containerB:80 LABEL_B

Maybe this is not the purpose of gobetween.
Many Thanks and Regards,
Javier R

• From config file
• From http response
• Consul, etc...

Formats: TOML, JSON.

with balance = "weight" everything works fine
but with balance = "roundrobin" gobetween won`t start.

please see config and error below:

[logging]
  level = "info"
  output = "/var/log/gobetween.log"

[api]
  enabled = true
  bind = "10.0.0.8:8888"

[defaults]
  max_connections = 0
  client_idle_timeout = "0"
  backend_idle_timeout = "0"
  backend_connection_timeout = "0"

[servers]
  [servers.es]
    max_connections = 10000
    client_idle_timeout = "10m"
    backend_idle_timeout = "10m"
    backend_connection_timeout = "2s"
    bind = "10.0.0.8:9200"
    protocol = "tcp"
    balance = "weight"
    [servers.es.discovery]
      kind = "static"
      failpolicy = "keeplast"
      interval = "0"
      timeout = "0"
      static_list = ["10.0.0.51:9200", "10.0.0.52:9200", "10.0.0.53:9200", "10.0.0.54:9200", "10.0.0.55:9200"]
    [servers.es.healthcheck]
      kind = "ping"
      interval = "1s"
      passes = 3
      fails = 3
      timeout = "500ms"

root@??:~# /usr/sbin/gobetween -c /etc/gobetween.toml
2016-07-10 14:36:47 [INFO ] (main): gobetween v0.2.0+dev
2016-07-10 14:36:47 [INFO ] (main): Using config file /etc/gobetween.toml
panic: interface conversion: *balance.RoundrobinBalancer is not balance.Balancer: missing method Elect

goroutine 6 [running]:
panic(0xa3b8e0, 0xc82012fe40)
    /usr/local/go/src/runtime/panic.go:464 +0x3e6
_/home/ubuntu/git_projects/gobetween/src/balance.New(0xc8201b4d80, 0xa, 0x0, 0x0)
    /home/ubuntu/git_projects/gobetween/src/balance/balancer.go:33 +0x17f
_/home/ubuntu/git_projects/gobetween/src/server.New(0xc8201b95c8, 0xd, 0xc8201b5bf0, 0xc8201b5bb0, 0xc8201c4240, 0xc8201c4260, 0xc8201b4c80, 0xd, 0xc8201b4d00, 0x3, ...)
    /home/ubuntu/git_projects/gobetween/src/server/server.go:74 +0x13a
_/home/ubuntu/git_projects/gobetween/src/manager.Create(0xc8201b95c8, 0xd, 0xc8201b5bf0, 0xc8201b5bb0, 0xc8201c4240, 0xc8201c4260, 0xc8201b4c80, 0xd, 0xc8201b4d00, 0x3, ...)
    /home/ubuntu/git_projects/gobetween/src/manager/manager.go:121 +0x160
_/home/ubuntu/git_projects/gobetween/src/manager.Initialize(0xc8201b4310, 0x4, 0xc8201b43d0, 0x7, 0x1, 0xc8201b4560, 0xd, 0xc8201b5a10, 0xc8201b59b0, 0xc8201b59d0, ...)
    /home/ubuntu/git_projects/gobetween/src/manager/manager.go:46 +0x27b
created by main.main
    /home/ubuntu/git_projects/gobetween/src/main.go:83 +0x4f3

During SRV discovery one or more services have empty port key in backend definition:

{
  "host": "172.16.160.67",
  "port": "",
    "priority": 0,
    "stats": {
      "active_connections": 0,
      "live": false,
      "rx": 0,
      "rx_second": 0,
      "total_connections": 0,
      "tx": 0,
      "tx_second": 0
    },
    "weight": 0
}

Manual lookup returns correct SRV records:

;; ANSWER SECTION:
nginx.service.consul.   0   IN  SRV 1 1 9802 m-slave-01.ua.int.node.ua.consul.
nginx.service.consul.   0   IN  SRV 1 1 26661 m-slave-02.ua.int.node.ua.consul.
nginx.service.consul.   0   IN  SRV 1 1 57048 m-slave-04.ua.int.node.ua.consul.

;; ADDITIONAL SECTION:
m-slave-01.ua.int.node.ua.consul. 0 IN  A   172.16.160.50
m-slave-02.ua.int.node.ua.consul. 0 IN  A   172.16.160.60
m-slave-04.ua.int.node.ua.consul. 0 IN  A   172.16.160.143

Allow to configure TLS intead of TCP while proxying traffic to backends.

Configuration options may be similar to nginx.

Add kubernetes discovery

Does this project support 1.12+ ? As there is no mention of it, I guess it's not.

I'm still looking for a solid reverse-proxy under Docker Swarm 1.12+ (using 1.12.1 at the moment). Here some [solutions](Does this project support 1.12+ ?) meanwhile.

Cheers!