generate_queries.rb

#!/usr/bin/env ruby

def main
  if ARGV.length < 2
    $stderr.puts "Usage: #{$PROGRAM_NAME} <query-tmpl> <max-id> " \
                 "[range-size] [range-begin] [sleep]"
    exit 1
  end

  query_tmpl = File.read(ARGV.fetch(0))
  max_id = Integer(ARGV.fetch(1))
  range_size = Integer(ARGV.fetch(2, 10_000))
  range_begin = Integer(ARGV.fetch(3, 0))
  min_id = range_begin
  sleeps = Float(ARGV.fetch(4, 1))

  range_end = range_begin + range_size

  loop do
    if range_begin > max_id
      $stderr.puts '-- ALL DONE'
      break
    end

    range_end = max_id if range_end > max_id

    print_query(query_tmpl, range_begin, range_end, min_id, max_id, sleeps)

    range_begin += range_size
    range_end = range_begin + range_size
  end
end

def print_query(query_tmpl, range_begin, range_end, min_id, max_id, sleeps)
  $stderr.puts <<~EOF
    --
    -- range_begin=#{range_begin} range_end=#{range_end}
    --
  EOF

  $stdout.puts query_tmpl % {
    range_begin: range_begin,
    range_end: range_end - 1,
    min_id: min_id,
    max_id: max_id,
    sleeps: sleeps,
    pct_completed: (Float(range_end) / Float(max_id)) * 100.0
  }

  $stderr.puts
end

main if $PROGRAM_NAME == __FILE__

query-template.sql

BEGIN;
  INSERT INTO logs(
    id, job_id, content,
    removed_by, created_at, updated_at, aggregated_at, archived_at, purged_at,
    removed_at, archiving, archive_verified
  )
  SELECT
    id, job_id, NULL as content,
    removed_by, created_at, updated_at, aggregated_at, archived_at, purged_at,
    removed_at, false as archiving, true as archive_verified
  FROM logs_recovery
  WHERE id BETWEEN %{range_begin} AND %{range_end}
    AND id NOT IN (
      SELECT id FROM logs WHERE id BETWEEN %{range_begin} AND %{range_end}
    );
COMMIT;
COPY (SELECT '%{pct_completed}%% complete; sleeping %{sleeps}s') TO STDOUT;
SELECT now(), pg_sleep(%{sleeps});