Monitor Sidekiq Scheduled Jobs for Empty and Missed Runs

Sidekiq counts a job as processed when perform returns. Retries, the dead set, and the busy list tell you about exceptions. They say nothing about a job that ran, found nothing to do, and returned.

Scheduled jobs add a second gap. Whether you use sidekiq-cron, sidekiq-scheduler, or a plain cron entry that enqueues, a stopped scheduler produces no errors. The job is simply never enqueued.

Post a receipt from the job

Net::HTTP is enough. No gem is needed.

require "json"
require "net/http"
require "time"
require "uri"

begin
  uri = URI("https://app.crontinel.com/api/v1/ingest/cron")
  req = Net::HTTP::Post.new(uri)
  req["Authorization"] = "Bearer #{ENV.fetch("CRONTINEL_INGEST_KEY")}"
  req["Content-Type"] = "application/json"
  req.body = {
    request_key: "sidekiq-#{jid}-#{started}",
    command: "reports:generate",
    status: "completed",
    exit_code: 0,
    started_at: started,
    finished_at: Time.now.utc.iso8601,
    outcomes: { metrics: { processed_records: records } },
  }.to_json
  Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) { |http| http.request(req) }
rescue StandardError
  # Monitoring must not fail the business job.
end

Capture started = Time.now.utc.iso8601 at the top of perform, count what the job handled in records, and run the block in an ensure. The key joins jid and the start time. A retry keeps the jid but starts later, so it reports as its own run. A key reused after a run has finished is rejected, and a failed first attempt would otherwise block the successful retry from being recorded.

The rescue matters. If Crontinel is slow or down, the job must not start failing and retrying because of it.

Alert on a job that was never enqueued

Register the job with its schedule:

curl -sS -X POST "https://app.crontinel.com/api/v1/ingest/schedule" \
  -H "Authorization: Bearer $CRONTINEL_INGEST_KEY" \
  -H "Content-Type: application/json" \
  -d '{"source":"named_by_user","tasks":[{"command":"reports:generate","expression":"0 3 * * *","timezone":"UTC","grace_seconds":300,"counted_minimum":1}]}'

When no receipt arrives inside the grace window, Crontinel opens a “cron never started” alert. When a receipt arrives with fewer than one record, the outcome rule fails the run even though Sidekiq called it a success.

What was tested

The snippet ran with plain Ruby against a local Crontinel app and landed as a completed run. Running it outside Rails showed it needs require "time" for Time#iso8601, which the version above includes. It was not run inside a Sidekiq process.

See the check-in recipes for the other runtimes.