deadlock_test.py

from pulpcore.app.models.content import Artifact, ContentArtifact
from pulp_file.app.models import FileContent
from hashlib import sha256
from django.db import connection
from multiprocessing import Process
from time import sleep

artifacts = {}  # filename: Artifact
content = {}  # filename: FileContent
print(">>>BUILDING CONTENT/CA/ARTIFACTS...")
for i in range(20000):
  path = f'/tmp/{i:06d}.txt'
  content_path = f'{i:06d}.txt'
  with open(path, "w") as f:
    f.write(path)
  with open(path, "rb") as f:
    sum256 = sha256(f.read()).hexdigest()
  attrs = {"relative_path": content_path, "digest": sum256}
  fc = FileContent(**attrs)
  fc.save()
  content[path] = fc
  attrs = {"file": path, "sha256": sum256, "size": i}
  a = Artifact(**attrs)
  a.save()
  artifacts[path] = a

# 10 lists of ContentArtifacts
ca_lists = [[], [], [], [], []]
for k in content.keys():
  for i in range(5):
    attrs = {"content": fc, "relative_path": f'{i}/{content[k].relative_path}'}
    ca = ContentArtifact(**attrs)
    ca.save()
    ca.artifact = artifacts[k]
    ca_lists[i].append(ca)
    
def bulk_doit(ca_list):
  print(">>> ENTER...")
  connection.close()
  connection.connect()
  ContentArtifact.objects.bulk_update(ca_list, ["artifact"], batch_size=500)
  print(">>> EXIT...")


#Repeat a few  times to make sure the deadlock issue can be reproduced
for r in range(3):
  print("Try round: %s" % str(r + 1))
  # Ensure there are something to update when rerunning this part
  connection.connect()
  ContentArtifact.objects.filter().update(artifact_id=None)
  processes = []
  for i in range(5):
    for j in range(4):
      p = Process(target=bulk_doit, args=(ca_lists[0],))
      processes.append(p)
      p.start()
    sleep(3)
  for p in processes:
    p.join()