Files
mtusi2026lab4/scripts/plot_results.py
T

69 lines
5.0 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""Static research figures from recorded evidence only; no new measurements."""
import csv
import json
from pathlib import Path
import re
import statistics
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
ROOT=Path(__file__).resolve().parents[1]
OUT=ROOT/'evidence/figures';OUT.mkdir(exist_ok=True)
plt.rcParams.update({'font.family':'DejaVu Sans','font.size':10,'axes.spines.top':False,'axes.spines.right':False})
def save(fig,name):
fig.tight_layout()
fig.savefig(OUT/(name+'.png'),dpi=170)
fig.savefig(OUT/(name+'.svg'))
plt.close(fig)
def bands(ax,x,groups,label,color):
med=[statistics.median(v) for v in groups]
ax.errorbar(x,med,yerr=[[m-min(v) for m,v in zip(med,groups)],[max(v)-m for m,v in zip(med,groups)]],fmt='o-',capsize=4,label=label,color=color)
r=json.loads((ROOT/'evidence/reconcile.json').read_text(encoding='utf-8-sig'))
points=sorted({v['extra_objects'] for v in r['cost']})
groups=[[v for v in r['cost'] if v['extra_objects']==n] for n in points]
x=[v[0]['counts']['objects_scanned'] for v in groups]
fig,axes=plt.subplots(1,2,figsize=(10.5,4))
bands(axes[0],x,[[v['elapsed_seconds'] for v in vs] for vs in groups],'Медиана и min–max (3 повтора)','#286D8C')
axes[0].set(xlabel='Объектов в bucket',ylabel='Время dry-run, с',title='Стоимость полной сверки')
axes[0].legend(fontsize=8)
for field,label in [('head_requests','HEAD'),('get_requests','GET'),('list_requests','LIST')]:
axes[1].plot(x,[v[0]['counts'][field] for v in groups],'o-',label=label)
axes[1].set(xlabel='Объектов в bucket',ylabel='Число S3-запросов',title='Цена обхода: запросы')
axes[1].legend();save(fig,'reconcile-cost')
u=json.loads((ROOT/'evidence/upload-load.json').read_text(encoding='utf-8-sig'))
sizes=sorted({v['size_bytes'] for v in u['runs']}); x=[s/1048576 for s in sizes]
groups=[[v for v in u['runs'] if v['size_bytes']==s] for s in sizes]
fig,axes=plt.subplots(1,2,figsize=(10.5,4))
bands(axes[0],x,[[v['successful_rps'] for v in vs] for vs in groups],'3 партии × 10 клиентов','#286D8C')
axes[0].set(xlabel='PNG, MiB',ylabel='Успешных загрузок/с',title='Синхронный upload через API')
bands(axes[1],x,[[v['p95_seconds'] for v in vs] for vs in groups],'p95 успешных запросов','#B15E38')
axes[1].set(xlabel='PNG, MiB',ylabel='Задержка, с',title='p95 внутри партии (10 запросов)')
for ax in axes: ax.legend(fontsize=8)
save(fig,'upload-throughput')
def mib(value):
match=re.match(r'([0-9.]+)([A-Za-z]+)',value.strip())
amount=float(match[1]);unit=match[2]
return amount*{'B':1/1048576,'KiB':1/1024,'MiB':1,'GiB':1024,'kB':1000/1048576,'MB':1000000/1048576,'GB':1000000000/1048576}[unit]
fig,ax=plt.subplots(figsize=(10.5,4))
first=u['samples'][0]['time']
for name in sorted({c['Name'] for s in u['samples'] for c in s['containers'] if '-api-' in c['Name']}):
pairs=[(s['time']-first,mib(c['MemUsage'].split('/')[0])) for s in u['samples'] for c in s['containers'] if c['Name']==name]
ax.plot([a for a,b in pairs],[b for a,b in pairs],'o-',label=name)
ax.axhline(512,color='#B15E38',linestyle='--',label='Лимит одной API: 512 MiB')
ax.set(xlabel='Время от первой выборки, с',ylabel='Docker working set, MiB',title='Память API: прогрев, загрузки и проверка оригиналов')
ax.legend(fontsize=8);save(fig,'upload-memory')
with (ROOT/'evidence/upload-summary.csv').open('w',newline='',encoding='utf-8') as f:
writer=csv.DictWriter(f,fieldnames=list(u['runs'][0]));writer.writeheader();writer.writerows(u['runs'])
summary={'uploads':[], 'cost':[],'sampled_memory_max_mib':{},'kernel_memory_peak_mib':{}}
for size,vs in zip(sizes,groups):
summary['uploads'].append({'size_bytes':size,'successful_rps_median':statistics.median(v['successful_rps'] for v in vs),'p50_seconds_median':statistics.median(v['p50_seconds'] for v in vs),'p95_seconds_median':statistics.median(v['p95_seconds'] for v in vs),'p99_seconds_median':statistics.median(v['p99_seconds'] for v in vs),'errors':sum(v['errors'] for v in vs),'attempts':sum(v['attempts'] for v in vs),'elapsed_seconds':sum(v['elapsed_seconds'] for v in vs)})
for n in points:
vs=[v for v in r['cost'] if v['extra_objects']==n]
summary['cost'].append({'objects':vs[0]['counts']['objects_scanned'],'elapsed_median':statistics.median(v['elapsed_seconds'] for v in vs),'elapsed_min':min(v['elapsed_seconds'] for v in vs),'elapsed_max':max(v['elapsed_seconds'] for v in vs),'counts':vs[0]['counts']})
for s in u['samples']:
for c in s['containers']:
name=c['Name'];summary['sampled_memory_max_mib'][name]=max(summary['sampled_memory_max_mib'].get(name,0),mib(c['MemUsage'].split('/')[0]))
for c in u.get('cgroup_after',[]):summary['kernel_memory_peak_mib'][c['id']]=int(c['memory.peak'])/1048576
(ROOT/'evidence/summary.json').write_text(json.dumps(summary,indent=2),encoding='utf-8')
print(json.dumps(summary,indent=2))