We are working on migrating some servers which had sensu-core clients, to sensugo and then realised that both tries to use the same socket port 3030. Looking for some help to understand if this port is used in the normal pub-sub configurations, ie, getting the check results back to server.
No the socket port at 3030 is not used for normal use cases. In classic it was a way to send adhoc events generated from other processes… like from a cronjob for example, or initscript wrapper.
For the side by side transition I would recommend disabling the socket port in the new sensu-agent via the config option to disable it.
I should also add… the sensu-agent also exposes an event api on port 3031 which is sort of the new approach for the same purpose. As you grow into your agent usage, you may want to make use of that events api. Anyone who was using the socket port should transition to using the agent’s events api to ensure best operation across the transition to sensu-go.
For example (and I think I actually had a pattern for this or someone did) on linux systems running systemd you can define a systemd on failure hook script that could send a Sensu alert into the local sensu-agent’s event api, and the sensu-agent would handle all the auth back to the sensu-backend.