Cannot add retention policies via update to yaml config

[mhaagens]

I'm trying to add retention policies via YAML in my Docker image, but getting this error

** (exit) exited in: GenServer.call({:via, :syn, {:replication, Sequin.Runtime.WalEventSupervisor}}, {:start_child, {{Sequin.Runtime.WalPipelineServer, :start_link, [[replication_slot_id: "REDACTED", destination_oid: REDACTED, destination_database_id: "REDACTED", wal_pipeline_ids: ["REDACTED"]]]}, :temporary, 5000, :worker, [Sequin.Runtime.WalPipelineServer]}}, :infinity)
    ** (EXIT) no process: the process is not alive or there's no process currently associated with the given name, possibly because its application isn't started
    (elixir 1.18.4) lib/gen_server.ex:1121: GenServer.call/3
    (elixir 1.18.4) lib/enum.ex:992: anonymous fn/3 in Enum.each/2
    (stdlib 6.2.2.2) maps.erl:860: :maps.fold_1/4
    (elixir 1.18.4) lib/enum.ex:2558: Enum.each/2
    (sequin 0.1.0) lib/sequin/replication/replication.ex:302: Sequin.Replication.notify_wal_pipeline_changed/1
    (sequin 0.1.0) lib/sequin/replication/replication.ex:267: Sequin.Replication.create_wal_pipeline_with_lifecycle/2
    (sequin 0.1.0) lib/sequin/yaml_loader.ex:645: anonymous fn/4 in Sequin.YamlLoader.upsert_wal_pipelines/3
    nofile:1: (file)
[os_mon] cpu supervisor port (cpu_sup): Erlang has closed
[os_mon] memory supervisor port (memsup): Erlang has closed

[dosubot]

Hi @mhaagens! I'm Dosu and I’m helping the sequin team.

This looks like a race condition between YAML loading and the supervisor startup process.

When you configure retention policies via YAML in Docker, the YamlLoader starts the application to access the database, then creates WAL pipelines. During that process, notify_wal_pipeline_changed tries to call WalEventSupervisor via the :syn registry, but the Runtime.Supervisor may not have fully initialized and registered yet. This results in the :noproc error you're seeing.

Workaround: Configure retention policies via the UI or API after Sequin has fully started, rather than including them in the initial YAML config. You can then add them back to your YAML for subsequent deployments once the initial entities exist.

A similar race condition was reported in #2103 affecting other components during process lifecycle transitions.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}