prop/test/microwaveprop/workers/radar_frame_worker_test.exs
Graham McIntire ddedd241de
fix(radar): dedupe RadarFrameWorker jobs by frame_ts
BackfillEnqueueWorker re-enqueued every :queued contact each cycle, and
RadarFrameWorker had no unique clause, so the same 5-min NEXRAD frame
got scheduled repeatedly with overlapping contact lists. Prod had 9,134
duplicate available jobs across 1,061 distinct frames — 89% waste.

Two coupled changes so contacts aren't stranded by the dedup:

  1. Worker now queries eligible contacts by the frame's 5-min bucket
     instead of the args' contact_ids list. A contact that flips to
     :queued between the enqueue and the job running gets picked up.

  2. unique: [frame_ts] on states [available, scheduled, retryable].
     :executing intentionally excluded — a contact that flips to :queued
     mid-execution can still enqueue a follow-up instead of getting
     stuck behind a running job that missed it.
2026-04-20 14:21:55 -05:00

149 lines
5.4 KiB
Elixir

defmodule Microwaveprop.Workers.RadarFrameWorkerTest do
use Microwaveprop.DataCase, async: false
use Oban.Testing, repo: Microwaveprop.Repo
alias Microwaveprop.Radio.Contact
alias Microwaveprop.Radio.ContactCommonVolumeRadar
alias Microwaveprop.Repo
alias Microwaveprop.Weather.NexradClient
alias Microwaveprop.Workers.RadarFrameWorker
setup do
Req.Test.set_req_test_from_context(%{async: false})
:ok
end
defp insert_contact(attrs \\ %{}) do
default = %{
station1: "W5XD",
station2: "K5TR",
qso_timestamp: ~U[2024-09-15 18:30:00Z],
mode: "CW",
band: Decimal.new("10000"),
grid1: "EM12",
grid2: "EM00",
pos1: %{"lat" => 32.9, "lon" => -97.0},
pos2: %{"lat" => 33.1, "lon" => -96.8},
distance_km: Decimal.new("30")
}
{:ok, contact} =
%Contact{}
|> Contact.changeset(Map.merge(default, attrs))
|> Repo.insert()
contact
end
describe "process_frame/4" do
test "writes a ContactCommonVolumeRadar row and marks :complete for every contact in the batch" do
c1 = insert_contact(%{station1: "A1"})
c2 = insert_contact(%{station1: "A2"})
# In-memory pixel buffer: 10x10 empty frame. The aggregator will
# observe zero echoes but still emit a stats row per contact.
pixels = :binary.copy(<<0>>, 100)
width = 10
frame_ts = ~U[2024-09-15 18:30:00Z]
assert :ok =
RadarFrameWorker.process_frame(
[Repo.reload!(c1), Repo.reload!(c2)],
frame_ts,
pixels,
width
)
assert Repo.get_by(ContactCommonVolumeRadar, contact_id: c1.id)
assert Repo.get_by(ContactCommonVolumeRadar, contact_id: c2.id)
assert %Contact{radar_status: :complete} = Repo.get!(Contact, c1.id)
assert %Contact{radar_status: :complete} = Repo.get!(Contact, c2.id)
end
end
describe "perform/1" do
test "marks every contact in the batch :unavailable when the frame 404s" do
c1 = insert_contact(%{station1: "B1"})
c2 = insert_contact(%{station1: "B2"})
Req.Test.stub(NexradClient, fn conn ->
Plug.Conn.send_resp(conn, 404, "not found")
end)
assert :ok =
perform_job(RadarFrameWorker, %{
"frame_ts" => "2024-09-15T18:30:00Z",
"contact_ids" => [c1.id, c2.id]
})
refute Repo.get_by(ContactCommonVolumeRadar, contact_id: c1.id)
refute Repo.get_by(ContactCommonVolumeRadar, contact_id: c2.id)
assert %Contact{radar_status: :unavailable} = Repo.get!(Contact, c1.id)
assert %Contact{radar_status: :unavailable} = Repo.get!(Contact, c2.id)
end
test "skips contacts already at a terminal status (idempotent replay)" do
c_done = insert_contact(%{station1: "DONE"})
{:ok, _} = c_done |> Ecto.Changeset.change(%{radar_status: :complete}) |> Repo.update()
c_new = insert_contact(%{station1: "NEW"})
Req.Test.stub(NexradClient, fn conn ->
Plug.Conn.send_resp(conn, 404, "not found")
end)
assert :ok =
perform_job(RadarFrameWorker, %{
"frame_ts" => "2024-09-15T18:30:00Z",
"contact_ids" => [c_done.id, c_new.id]
})
# Already-complete contact stays complete; new one is marked unavailable.
assert %Contact{radar_status: :complete} = Repo.get!(Contact, c_done.id)
assert %Contact{radar_status: :unavailable} = Repo.get!(Contact, c_new.id)
end
test "processes every eligible contact in the frame's 5-min bucket, not just contact_ids" do
# Both contacts land in the 18:30 bucket (18:30 ≤ ts < 18:35).
c_passed = insert_contact(%{station1: "PASSED", qso_timestamp: ~U[2024-09-15 18:30:00Z]})
c_omitted = insert_contact(%{station1: "OMITTED", qso_timestamp: ~U[2024-09-15 18:34:00Z]})
# Outside the 18:30 bucket — must not be touched.
c_other_bucket = insert_contact(%{station1: "OTHER", qso_timestamp: ~U[2024-09-15 18:35:00Z]})
Req.Test.stub(NexradClient, fn conn ->
Plug.Conn.send_resp(conn, 404, "not found")
end)
# Only c_passed is explicitly listed, but c_omitted is in the same
# bucket and still :pending — the worker must pick it up too.
assert :ok =
perform_job(RadarFrameWorker, %{
"frame_ts" => "2024-09-15T18:30:00Z",
"contact_ids" => [c_passed.id]
})
assert %Contact{radar_status: :unavailable} = Repo.get!(Contact, c_passed.id)
assert %Contact{radar_status: :unavailable} = Repo.get!(Contact, c_omitted.id)
# Outside-bucket contact stays :pending.
assert %Contact{radar_status: :pending} = Repo.get!(Contact, c_other_bucket.id)
end
end
describe "unique constraint" do
test "rejects a second insert with the same frame_ts while one is pending" do
frame_ts = "2024-09-15T18:30:00Z"
{:ok, first} =
Oban.insert(RadarFrameWorker.new(%{"frame_ts" => frame_ts, "contact_ids" => ["a"]}))
# Second insert with a different contact_ids list but same frame_ts
# must collapse into the existing job — otherwise backfill sweeps
# duplicate 5 MB PNG fetches per frame.
{:ok, second} =
Oban.insert(RadarFrameWorker.new(%{"frame_ts" => frame_ts, "contact_ids" => ["b"]}))
assert first.id == second.id
end
end
end