mirror of
https://github.com/maybe-finance/maybe.git
synced 2025-07-19 13:19:39 +02:00
Benchmarking setup (#2366)
* Benchmarking setup * Get demo data working in benchmark scenario * Finalize default demo scenario * Finalize benchmarking setup
This commit is contained in:
parent
cdad31812a
commit
84b2426e54
23 changed files with 1477 additions and 2166 deletions
154
lib/tasks/benchmarking.rake
Normal file
154
lib/tasks/benchmarking.rake
Normal file
|
@ -0,0 +1,154 @@
|
|||
# Benchmarking requires a production-like data sample, so requires some up-front setup.
|
||||
#
|
||||
# 1. Load a scrubbed production-like slice of data into maybe_benchmarking DB locally
|
||||
# 2. Setup .env.production so that the Rails app can boot with RAILS_ENV=production and connect to local maybe_benchmarking DB
|
||||
# 3. Run `rake benchmark_dump:06_setup_bench_user`
|
||||
# 4. Run locally, find endpoint needed
|
||||
# 5. Run an endpoint, example: `ENDPOINT=/budgets/jun-2025/budget_categories/245637cb-129f-4612-b0a8-1de57559372b RAILS_ENV=production BENCHMARKING_ENABLED=true RAILS_LOG_LEVEL=debug rake benchmarking:ips`
|
||||
namespace :benchmarking do
|
||||
# When to use: Track overall endpoint speed improvements over time (recommended, most practical test)
|
||||
desc "Run cold & warm performance benchmarks and append to history"
|
||||
task ips: :environment do
|
||||
path = ENV.fetch("ENDPOINT", "/")
|
||||
|
||||
# 🚫 Fail fast unless the benchmark is run in production mode
|
||||
unless Rails.env.production?
|
||||
raise "benchmark:ips must be run with RAILS_ENV=production (current: #{Rails.env})"
|
||||
end
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Tunable parameters – override with environment variables if needed
|
||||
# ---------------------------------------------------------------------------
|
||||
cold_warmup = Integer(ENV.fetch("COLD_WARMUP", 0)) # seconds to warm up before *cold* timing (0 == true cold)
|
||||
cold_iterations = Integer(ENV.fetch("COLD_ITERATIONS", 1)) # requests to measure for the cold run
|
||||
|
||||
warm_warmup = Integer(ENV.fetch("WARM_WARMUP", 5)) # seconds benchmark-ips uses to stabilise JIT/caches
|
||||
warm_time = Integer(ENV.fetch("WARM_TIME", 30)) # seconds benchmark-ips samples for warm statistics
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
setup_benchmark_env(path)
|
||||
FileUtils.mkdir_p("tmp/benchmarks")
|
||||
|
||||
timestamp = Time.current.strftime("%Y-%m-%d %H:%M:%S")
|
||||
commit_sha = `git rev-parse --short HEAD 2>/dev/null`.strip rescue "unknown"
|
||||
puts "🕒 Starting benchmark run at #{timestamp} (#{commit_sha})"
|
||||
|
||||
# 🚿 Flush application caches so the first request is a *true* cold hit
|
||||
Rails.cache&.clear if defined?(Rails)
|
||||
|
||||
# ---------------------------
|
||||
# 1️⃣ Cold measurement
|
||||
# ---------------------------
|
||||
puts "❄️ Running cold benchmark for #{path} (#{cold_iterations} iteration)..."
|
||||
cold_cmd = "IPS_WARMUP=#{cold_warmup} IPS_TIME=0 IPS_ITERATIONS=#{cold_iterations} " \
|
||||
"bundle exec derailed exec perf:ips"
|
||||
cold_output = `#{cold_cmd} 2>&1`
|
||||
|
||||
puts "Cold output:"
|
||||
puts cold_output
|
||||
|
||||
cold_result = extract_clean_results(cold_output)
|
||||
|
||||
# ---------------------------
|
||||
# 2️⃣ Warm measurement
|
||||
# ---------------------------
|
||||
puts "🔥 Running warm benchmark for #{path} (#{warm_time}s sample)..."
|
||||
warm_cmd = "IPS_WARMUP=#{warm_warmup} IPS_TIME=#{warm_time} " \
|
||||
"bundle exec derailed exec perf:ips"
|
||||
warm_output = `#{warm_cmd} 2>&1`
|
||||
|
||||
puts "Warm output:"
|
||||
puts warm_output
|
||||
|
||||
warm_result = extract_clean_results(warm_output)
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Persist results
|
||||
# ---------------------------------------------------------------------------
|
||||
separator = "\n" + "=" * 70 + "\n"
|
||||
timestamp_header = "#{separator}📊 BENCHMARK RUN - #{timestamp} (#{commit_sha})#{separator}"
|
||||
|
||||
# Table header
|
||||
table_header = "| Type | IPS | Deviation | Time/Iteration | Iterations | Total Time |\n"
|
||||
table_separator = "|------|-----|-----------|----------------|------------|------------|\n"
|
||||
|
||||
cold_row = format_table_row("COLD", cold_result)
|
||||
warm_row = format_table_row("WARM", warm_result)
|
||||
|
||||
combined_result = table_header + table_separator + cold_row + warm_row + "\n"
|
||||
|
||||
File.open(benchmark_file(path), "a") { |f| f.write(timestamp_header + combined_result) }
|
||||
|
||||
puts "✅ Results saved to #{benchmark_file(path)}"
|
||||
end
|
||||
|
||||
private
|
||||
def setup_benchmark_env(path)
|
||||
ENV["USE_AUTH"] = "true"
|
||||
ENV["USE_SERVER"] = "puma"
|
||||
ENV["PATH_TO_HIT"] = path
|
||||
ENV["HTTP_METHOD"] = "GET"
|
||||
ENV["RAILS_LOG_LEVEL"] ||= "error" # keep output clean
|
||||
end
|
||||
|
||||
def benchmark_file(path)
|
||||
filename = case path
|
||||
when "/" then "dashboard"
|
||||
else
|
||||
path.gsub("/", "_").gsub(/^_+/, "")
|
||||
end
|
||||
"tmp/benchmarks/#{filename}.txt"
|
||||
end
|
||||
|
||||
def extract_clean_results(output)
|
||||
lines = output.split("\n")
|
||||
|
||||
# Example benchmark-ips output line:
|
||||
# " SomeLabel 14.416k (± 3.8%) i/s - 72.000k in 5.004618s"
|
||||
result_line = lines.find { |line| line.match(/\d[\d\.kM]*\s+\(±\s*[0-9\.]+%\)\s+i\/s/) }
|
||||
|
||||
if result_line
|
||||
if (match = result_line.match(/(\d[\d\.kM]*)\s+\(±\s*([0-9\.]+)%\)\s+i\/s\s+(?:\(([^)]+)\)\s+)?-\s+(\d[\d\.kM]*)\s+in\s+(\d+\.\d+)s/))
|
||||
ips_value = match[1]
|
||||
deviation_percent = match[2].to_f
|
||||
time_per_iteration = match[3] || "-"
|
||||
iterations = match[4]
|
||||
total_time = "#{match[5]}s"
|
||||
|
||||
{
|
||||
ips: ips_value,
|
||||
deviation: "± %.2f%%" % deviation_percent,
|
||||
time_per_iteration: time_per_iteration,
|
||||
iterations: iterations,
|
||||
total_time: total_time
|
||||
}
|
||||
else
|
||||
no_data_hash
|
||||
end
|
||||
else
|
||||
no_data_hash("No results")
|
||||
end
|
||||
end
|
||||
|
||||
def format_table_row(type, data)
|
||||
# Wider deviation column accommodates strings like "± 0.12%"
|
||||
"| %-4s | %-5s | %-11s | %-14s | %-10s | %-10s |\n" % [
|
||||
type,
|
||||
data[:ips],
|
||||
data[:deviation],
|
||||
data[:time_per_iteration],
|
||||
data[:iterations],
|
||||
data[:total_time]
|
||||
]
|
||||
end
|
||||
|
||||
def no_data_hash(ips_msg = "No data")
|
||||
{
|
||||
ips: ips_msg,
|
||||
deviation: "-",
|
||||
time_per_iteration: "-",
|
||||
iterations: "-",
|
||||
total_time: "-"
|
||||
}
|
||||
end
|
||||
end
|
|
@ -1,39 +1,63 @@
|
|||
namespace :demo_data do
|
||||
desc "Creates a new user with no data. Use for testing empty data states."
|
||||
desc "Load empty demo dataset (no financial data)"
|
||||
task empty: :environment do
|
||||
families = [ "Demo Family 1" ]
|
||||
Demo::Generator.new.reset_and_clear_data!(families)
|
||||
start = Time.now
|
||||
puts "🚀 Loading EMPTY demo data…"
|
||||
|
||||
Demo::Generator.new.generate_empty_data!
|
||||
|
||||
puts "✅ Done in #{(Time.now - start).round(2)}s"
|
||||
end
|
||||
|
||||
desc "Creates a new user who has to go through onboarding still. Use for testing onboarding flows."
|
||||
desc "Load new-user demo dataset (family created but not onboarded)"
|
||||
task new_user: :environment do
|
||||
families = [ "Demo Family 1" ]
|
||||
Demo::Generator.new.reset_and_clear_data!(families, require_onboarding: true)
|
||||
start = Time.now
|
||||
puts "🚀 Loading NEW-USER demo data…"
|
||||
|
||||
Demo::Generator.new.generate_new_user_data!
|
||||
|
||||
puts "✅ Done in #{(Time.now - start).round(2)}s"
|
||||
end
|
||||
|
||||
desc "General data reset that loads semi-realistic data"
|
||||
task :reset, [ :count ] => :environment do |t, args|
|
||||
count = (args[:count] || 1).to_i
|
||||
families = count.times.map { |i| "Demo Family #{i + 1}" }
|
||||
Demo::Generator.new.reset_data!(families)
|
||||
desc "Load full realistic demo dataset"
|
||||
task default: :environment do
|
||||
start = Time.now
|
||||
seed = ENV.fetch("SEED", Random.new_seed)
|
||||
puts "🚀 Loading FULL demo data (seed=#{seed})…"
|
||||
|
||||
generator = Demo::Generator.new(seed: seed)
|
||||
generator.generate_default_data!
|
||||
|
||||
validate_demo_data!
|
||||
|
||||
elapsed = Time.now - start
|
||||
puts "🎉 Demo data ready in #{elapsed.round(2)}s"
|
||||
end
|
||||
|
||||
desc "Use this when you need to test multi-currency features of the app with a minimal setup"
|
||||
task multi_currency: :environment do
|
||||
families = [ "Demo Family 1", "Demo Family 2" ]
|
||||
Demo::Generator.new.generate_multi_currency_data!(families)
|
||||
end
|
||||
# ---------------------------------------------------------------------------
|
||||
# Validation helpers
|
||||
# ---------------------------------------------------------------------------
|
||||
def validate_demo_data!
|
||||
total_entries = Entry.count
|
||||
trade_entries = Entry.where(entryable_type: "Trade").count
|
||||
categorized_txn = Transaction.joins(:category).count
|
||||
txn_total = Transaction.count
|
||||
|
||||
desc "Use this when you want realistic budget data"
|
||||
task basic_budget: :environment do
|
||||
families = [ "Demo Family 1" ]
|
||||
Demo::Generator.new.generate_basic_budget_data!(families)
|
||||
end
|
||||
coverage = ((categorized_txn.to_f / txn_total) * 100).round(1)
|
||||
|
||||
# DO NOT RUN THIS unless you're testing performance locally. It will take a long time to load/clear. Easiest to clear with a db:reset
|
||||
desc "Generates realistic data for 500 families for performance testing. Creates 1 family with Ruby, then efficiently duplicates it 499 times using SQL bulk operations."
|
||||
task performance_testing: :environment do
|
||||
families = [ "Performance Family 1" ]
|
||||
Demo::Generator.new.generate_performance_testing_data!(families)
|
||||
puts "\n📊 Validation Summary".ljust(40, "-")
|
||||
puts "Entries total: #{total_entries}"
|
||||
puts "Trade entries: #{trade_entries} (#{trade_entries.between?(500, 1000) ? '✅' : '❌'})"
|
||||
puts "Txn categorization: #{coverage}% (>=75% ✅)"
|
||||
|
||||
unless total_entries.between?(8_000, 12_000)
|
||||
raise "Total entries #{total_entries} outside 8k–12k range"
|
||||
end
|
||||
unless trade_entries.between?(500, 1000)
|
||||
raise "Trade entries #{trade_entries} outside 500–1 000 range"
|
||||
end
|
||||
unless coverage >= 75
|
||||
raise "Categorization coverage below 75%"
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue