Hi!
One of my services must be running in 4 replicas on each of 16 servers.
And number of replicas may increase ( as a number of servers)
Creating deployment steps for each replica is difficult in this situation.
I dont know how can i optimize this process.
Can you add new feature?
For example, we can create step with powershell script. It will create number of variables (such as installation folder and service name) and run next step from the cycle.
Thanks for reaching out. Unfortunately it wouldnt be possible to add such request as there is no way to tell Octopus to run a specific step from another step. Octopus will run all the steps based on the snapshot of the deployment process that it took when the release was created.
The dynamic amount of servers should not be a problem, as you can run your steps based on server roles, and then Octopus will execute those steps on all the servers with that one on that environment. Regarding your replicas:
Are all replicas on each server running the exact same code( same binaries, same configuration)? What’s the difference between each replica on a server (ie different directory)?
If you can provide us more info about your scenario, we might be able to help you setting up the process.
All replicas on the same node are running same program code.
Main difference is in configuration files. Each replica of service must have uniq connection string, for example.
In this way i must create 64 roles.
Just to double check before coming up with a strategy: In total you’d have 4 unique connection strings (each server has the same unique 4), or 64 (each replica on each server is different)?