High Availability

[irtaza9]

I want to achieve the HA for my services. Is there any way to achieve this by using dokploy?

[Siumauricio]

Hi, yes you can achieve HA using dokploy, however you need a external load balancer, probably in the next weeks i will make a guide, people on discord asked the same, but yes is possible

[irtaza9]

@Siumauricio any update on this?

[ravinataraju]

I have currently migrated to cloudflare dns and have setup load balancer. I have setup an application (3 replicas) pointing to test domain which is load balanced by cloudflare, one ip mapped to dokploy installed vps for which health monitor is ok, but the ip of the secondary and third master health monitor is not working. The idea is for any reason the first master is down, docker swarm will take over on either secondary or third master so the internet traffic will still get a response. I might be wrong and looking forward to the HA documentation to do it properly. I kindly request to make this a high priority

[kaveh-ahmadian]

@Siumauricio was the guide ever published?

[jameshewitt1399]

@Siumauricio any update?

[jan-paulus]

@Siumauricio any update on this? Would be great if this was documented more.

[ravinataraju]

It would be beneficial to add the usage of swarm and remote server in HA, also a check box to setup replica database whilst setting up the database either within the swarm or remote servers whichever is the correct way to do it

[sunnycatnice]

Implementing a complete HA (High Availability) solution, as mentioned before, including shared storage across machines—even for the database, not just the application—would be ideal. Anyway I encountered issues because setting up HA with Docker Swarm for my volumes is quite challenging, and there doesn’t seem to be a "standardized" approach for achieving this.

[Siumauricio]

Nothing restricts you to have HA, I currently have HA for dokploy cloud and it works fine, just implement an external load balancer and that would be enough

[voale-rogii]

Hello team, you tell us about HA, but yours script installs traefik only one node.
Lets suppose there is a haproxy to node1, node2, node3 with healthcheks and there is an A dns record to haproxy host.
I'm installing dokploy on node1 (traefik with rules installed only here) and then adding 2 master nodes in swarm mode.
Then suppose node1 is making unavailable.
Describe us please where is HA here?

[BobbyMoore]

@voale-rogii - as far as I understand you would need Traefik on an external/separate host

[4volodin]

"Traefik on an external/separate host" - it's not the best case because it's need setup a tcp access to the docker on master nodes(for traefik magic)
I think the best case is:

• self install docker swarm with min 3 master nodes
• self install traefik
  with

--providers.docker=true
--providers.docker.swarmMode=true

..

deploy:
      mode: global
      placement:
        constraints:
          - node.role == manager

• tls termination on external host with vrrp or using for example AWS ALB
  sure, you can terminate tls on the traefik + consul for storing certs without external host

[4volodin]

but I don't know how will dokploy work with 3 traefik replicas ?

consul https://doc.traefik.io/traefik/v1.7/user-guide/cluster/