YunoHost/check-http
1
0
Fork 0
mirror of https://github.com/YunoHost/check-http.git synced 2024-09-03 19:56:42 +02:00
Code Issues Projects Releases Packages Wiki Activity

Merge pull request #1 from YunoHost/check-http-with-nounce-and-check-ports

Browse source 
Check http with nounce + add check ports feature
This commit is contained in:
Bram 2020-12-17 03:22:17 +01:00 committed by GitHub
commit f5e9385955
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 380 additions and 206 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

206
server.py
View file

@ -1,206 +0,0 @@
import time
import asyncio
import aiodns
import aiohttp
import validators
from sanic import Sanic
from sanic.log import logger
from sanic.response import html, json as json_response
from sanic.exceptions import InvalidUsage
app = Sanic()
# keep that in memory
RATE_LIMIT_DB = {}
# to prevent DDoS or bounce attack attempt or something like that
RATE_LIMIT_SECONDS = 5
def clear_rate_limit_db(now):
to_delete = []
"Remove too old rate limit values"
for key, value in RATE_LIMIT_DB.items():
if now - value > RATE_LIMIT_SECONDS:
# a dictionnary can't be modified during iteration so delegate this
# operation
to_delete.append(key)
for key in to_delete:
del RATE_LIMIT_DB[key]
async def query_dns(host, dns_entry_type):
loop = asyncio.get_event_loop()
dns_resolver = aiodns.DNSResolver(loop=loop)
try:
return await dns_resolver.query(host, dns_entry_type)
except aiodns.error.DNSError:
return []
except Exception:
import traceback
traceback.print_exc()
logger.error("Unhandled error while resolving DNS entry")
@app.route("/check/", methods=["POST"])
async def check_http(request):
"""
This function received an HTTP request from a YunoHost instance while this
server is hosted on our infrastructure. The expected request body is:
{"domain": "domain-to-check.tld"} and the method POST
The general workflow is the following:
- grab the ip from the request
- check for ip based rate limit (see RATE_LIMIT_SECONDS value)
- get json from body and domain from it
- check for domain based rate limit (see RATE_LIMIT_SECONDS value)
- check domain is in valid format
- check dns entry for domain match the ip of the request (advanced rule for ipv6)
- everything is checked, now try to do an http request on the domain
- answer saying if the domain can be reached
"""
# this is supposed to be a fast operation if run enough
now = time.time()
clear_rate_limit_db(now)
ip = request.ip
if ip in RATE_LIMIT_DB:
since_last_attempt = now - RATE_LIMIT_DB[ip]
if since_last_attempt < RATE_LIMIT_SECONDS:
logger.info(f"Rate limite {ip}, can retry in {int(RATE_LIMIT_SECONDS - since_last_attempt)} seconds")
return json_response({
"status": "error",
"code": "error_rate_limit",
"content": f"Rate limit on ip, retry in {int(RATE_LIMIT_SECONDS - since_last_attempt)} seconds",
}, status=400)
RATE_LIMIT_DB[ip] = time.time()
try:
data = request.json
except InvalidUsage:
logger.info(f"Unvalid json in request, body is : {request.body}")
return json_response({
"status": "error",
"code": "error_bad_json",
"content": "InvalidUsage, body isn't proper json",
}, status=400)
if not data or "domain" not in data:
logger.info(f"Unvalid request didn't specified a domain (body is : {request.body}")
return json_response({
"status": "error",
"code": "error_no_domain",
"content": "request must specify a domain",
}, status=400)
domain = data["domain"]
if domain in RATE_LIMIT_DB:
since_last_attempt = now - RATE_LIMIT_DB[domain]
if since_last_attempt < RATE_LIMIT_SECONDS:
logger.info(f"Rate limite {domain}, can retry in {int(RATE_LIMIT_SECONDS - since_last_attempt)} seconds")
return json_response({
"status": "error",
"code": "error_rate_limit",
"content": f"Rate limit on domain, retry in {int(RATE_LIMIT_SECONDS - since_last_attempt)} seconds",
}, status=400)
RATE_LIMIT_DB[domain] = time.time()
if not validators.domain(domain):
logger.info(f"Invalid request, is not in the right format (domain is : {domain})")
return json_response({
"status": "error",
"code": "error_domain_bad_format",
"content": "domain is not in the right format (do not include http:// or https://)",
}, status=400)
# TODO handle ipv6
# ipv6 situation
if ":" in ip:
dns_entry = await query_dns(domain, "AAAA")
if not dns_entry:
# check if entry in ip4 for custom error
dns_entry = await query_dns(domain, "A")
# there is an ipv4 entry but the request is made in ipv6, ask to uses ipv4 instead
if dns_entry:
logger.info(f"[ipv6] Invalid request, no AAAA DNS entry for domain {domain} BUT ipv4 entry, ask user to request in ipv4")
return json_response({
"status": "error",
"code": "error_no_ipv6_dns_entry_but_ipv4_dns_entry",
"content": f"there is not AAAA (ipv6) DNS entry for domain {domain} BUT there is an entry in ipv4, please redo the request in ipv4",
}, status=400)
else:
logger.info(f"[ipv6] Invalid request, no DNS entry for domain {domain} (both in ipv6 and ip4)")
return json_response({
"status": "error",
"code": "error_no_ipv4_ipv6_dns_entry_for_domain",
"content": f"there is not A (ipv4) and AAAA (ipv6) DNS entry for domain {domain}",
}, status=400)
# ipv4 situation
else:
dns_entry = await query_dns(domain, "A")
if not dns_entry:
logger.info(f"[ipv4] Invalid request, no DNS entry for domain {domain}")
return json_response({
"status": "error",
"code": "error_no_ipv4_dns_entry_for_domain",
"content": f"there is not A (ipv4) and AAAA (ipv6) DNS entry for domain {domain}",
}, status=400)
dns_entry = dns_entry[0]
if dns_entry.host != ip:
logger.info(f"Invalid request, A DNS entry {dns_entry.host} for domain {domain} doesn't match request ip {ip}")
return json_response({
"status": "error",
"code": "error_dns_entry_doesnt_match_request_ip",
"content": f"error, the request is made from the ip {ip} but the dns entry said {domain} has the ip {dns_entry.host}, you can only check a domain configured for your ip",
}, status=400)
async with aiohttp.ClientSession() as session:
try:
async with session.get("http://" + domain, timeout=aiohttp.ClientTimeout(total=30)) as response:
# XXX in the futur try to do a double check with the server to
# see if the correct content is get
await response.text()
logger.info(f"Success when checking http access for {domain} asked by {ip}")
# TODO various kind of errors
except aiohttp.client_exceptions.ClientConnectorError:
return json_response({
"status": "error",
"code": "error_http_check_connection_error",
"content": "connection error, could not connect to the requested domain, it's very likely unreachable",
}, status=418)
except Exception:
import traceback
traceback.print_exc()
return json_response({
"status": "error",
"code": "error_http_check_unknown_error",
"content": "an error happen while trying to get your domain, it's very likely unreachable",
}, status=400)
return json_response({"status": "ok"})
@app.route("/")
async def main(request):
return html("You aren't really supposed to use this website using your browser.<br><br>It's a small server to check if a YunoHost instance can be reached by http before trying to instal a LE certificate.")
if __name__ == "__main__":
app.run(host="0.0.0.0", port=7000)

369
yunodiagnoser.py Normal file
View file

@ -0,0 +1,369 @@
import re
import time
import asyncio
import aiohttp
import validators
import socket
from sanic import Sanic
from sanic.log import logger
from sanic.response import html, json as json_response
from sanic.exceptions import InvalidUsage
app = Sanic()
# ########################################################################### #
# Rate limit #
# ########################################################################### #
# keep that in memory
RATE_LIMIT_DB = {}
# to prevent DDoS or bounce attack attempt or something like that
# Can't do more than 10 requests in a 300-seconds window
RATE_LIMIT_SECONDS = 300
RATE_LIMIT_NB_REQUESTS = 10
def clear_rate_limit_db(now):
to_delete = []
"Remove too old rate limit values"
for key, times in RATE_LIMIT_DB.items():
# Remove values older RATE_LIMIT_SECONDS
RATE_LIMIT_DB[key] = [t for t in times if now - t < RATE_LIMIT_SECONDS]
# If list is empty, remove the key
if RATE_LIMIT_DB[key] == []:
# a dictionnary can't be modified during iteration so delegate this
# operation
to_delete.append(key)
for key in to_delete:
del RATE_LIMIT_DB[key]
def check_rate_limit(key, now):
# If there are more recent attempts than allowed
if key in RATE_LIMIT_DB and len(RATE_LIMIT_DB[key]) > RATE_LIMIT_NB_REQUESTS:
oldest_attempt = RATE_LIMIT_DB[key][0]
logger.info(f"Rate limit reached for {key}, can retry in {int(RATE_LIMIT_SECONDS - now + oldest_attempt)} seconds")
return json_response({
"error": {
"code": "error_rate_limit",
"content": f"Rate limit reached for this domain or ip, retry in {int(RATE_LIMIT_SECONDS - now + oldest_attempt)} seconds"
}
}, status=400)
# In any case, add this attempt to the DB
if key not in RATE_LIMIT_DB:
RATE_LIMIT_DB[key] = [now]
else:
RATE_LIMIT_DB[key].append(now)
# ########################################################################### #
# HTTP check #
# ########################################################################### #
@app.route("/check-http", methods=["POST"])
async def check_http(request):
"""
This function received an HTTP request from a YunoHost instance while this
server is hosted on our infrastructure. The request is expected to be a
POST request with a body like {"domains": ["domain1.tld", "domain2.tld"],
"nonce": "1234567890abcdef" }
The nonce value is a single-use ID, and we will try to reach
http://domain.tld/.well-known/ynh-{nonce} which should return 200 if we
are indeed reaching the right server.
The general workflow is the following:
- grab the ip from the request
- check for ip based rate limit (see RATE_LIMIT_SECONDS value)
- get json from body and domain from it
- check for domain-based rate limit (see RATE_LIMIT_SECONDS value)
- check domains are in valid format
- for each domain:
- try to do an http request on the ip (using the domain as target host) for the page /.well-known/ynh-diagnosis/{nonce}
- answer saying if the domain can be reached
"""
# this is supposed to be a fast operation if run often enough
now = time.time()
clear_rate_limit_db(now)
# ############################################# #
# Validate request and extract the parameters #
# ############################################# #
ip = request.headers["x-forwarded-for"].split(",")[0]
check_rate_limit_ip = check_rate_limit(ip, now)
if check_rate_limit_ip:
return check_rate_limit_ip
try:
data = request.json
except InvalidUsage:
logger.info(f"Invalid json in request, body is: {request.body}")
return json_response({
"error": {
"code": "error_bad_json",
"content": "Invalid usage, body isn't proper json"
}
}, status=400)
try:
assert data, "Empty request body"
assert isinstance(data, dict), "Request body ain't a proper dict"
assert "domains" in data, "No 'domains' provided"
assert "nonce" in data, "No 'nonce' provided"
# Check domain list format
assert isinstance(data["domains"], list), "'domains' ain't a list"
assert len(data["domains"]) > 0, "'domains' list is empty"
assert len(data["domains"]) < 30, "You cannot test that many domains"
for domain in data["domains"]:
assert isinstance(domain, str), "domain names must be strings"
assert len(domain) < 100, "Domain %s name seems pretty long, that's suspicious...?" % domain
assert len(data["domains"]) == len(set(data["domains"])), "'domains' list should contain unique elements"
# Check domain rate limit
for domain in data["domains"]:
check_rate_limit_domain = check_rate_limit(domain, now)
if check_rate_limit_domain:
return check_rate_limit_domain
# Check domains are valid domain names
for domain in data["domains"]:
assert validators.domain(domain), f"{domain} is not a valid domain"
# Check nonce format
assert isinstance(data["nonce"], str), "'nonce' ain't a string"
assert re.match(r"^[a-f0-9]{16}$", data["nonce"]), "'nonce' is not in the right forwat (it should be a 16-digit hexadecimal string)"
except AssertionError as e:
logger.info(f"Invalid request: {e} ... Original request body was: {request.body}")
return json_response({
"error": {
"code": "error_bad_json_data",
"content": f"Invalid request: {e} ... Original request body was: {request.body}"
}
}, status=400)
domains = data["domains"]
nonce = data["nonce"]
return json_response({
"http": {domain: await check_http_domain(ip, domain, nonce) for domain in domains}
})
async def check_http_domain(ip, domain, nonce):
if ":" in ip:
ip = "[%s]" % ip
async with aiohttp.ClientSession() as session:
try:
url = "http://" + ip + "/.well-known/ynh-diagnosis/" + nonce
async with session.get(url,
headers={"Host": domain},
allow_redirects=False,
timeout=aiohttp.ClientTimeout(total=5)) as response:
# XXX in the futur try to do a double check with the server to
# see if the correct content is get
await response.text()
# TODO various kind of errors
except (aiohttp.client_exceptions.ServerTimeoutError, asyncio.TimeoutError):
return {
"status": "error_http_check_timeout",
"content": "Timed-out while trying to contact your server from outside. It appears to be unreachable. You should check that you're correctly forwarding port 80, that nginx is running, and that a firewall is not interfering.",
}
except aiohttp.client_exceptions.ClientConnectorError as e:
return {
"status": "error_http_check_connection_error",
"content": "Connection error: could not connect to the requested domain, it's very likely unreachable. Raw error: " + str(e),
}
except Exception as e:
import traceback
traceback.print_exc()
return {
"status": "error_http_check_unknown_error",
"content": "An error happened while trying to reach your domain, it's very likely unreachable. Raw error: %s" % e,
}
if response.status != 200:
return {
"status": "error_http_check_bad_status_code",
"content": "Could not reach your server as expected, it returned code %s. It might be that another machine answered instead of your server. You should check that you're correctly forwarding port 80, that your nginx configuration is up to date, and that a reverse-proxy is not interfering." % response.status,
}
else:
return {
"status": "ok"
}
# ########################################################################### #
# Ports check #
# ########################################################################### #
@app.route("/check-ports/", methods=["POST"])
async def check_ports(request):
"""
This function received an HTTP request from a YunoHost instance while this
server is hosted on our infrastructure. The request is expected to be a
POST request with a body like {"ports": [80,443,22,25]}
The general workflow is the following:
- grab the ip from the request
- check for ip based rate limit (see RATE_LIMIT_SECONDS value)
- get json from body and ports list from it
- check ports are opened or closed
- answer the list of opened / closed ports
"""
# this is supposed to be a fast operation if run often enough
now = time.time()
clear_rate_limit_db(now)
# ############################################# #
# Validate request and extract the parameters #
# ############################################# #
ip = request.headers["x-forwarded-for"].split(",")[0]
check_rate_limit_ip = check_rate_limit(ip, now)
if check_rate_limit_ip:
return check_rate_limit_ip
try:
data = request.json
except InvalidUsage:
logger.info(f"Invalid json in request, body is: {request.body}")
return json_response({
"error": {
"code": "error_bad_json",
"content": "Invalid usage, body isn't proper json"
}
}, status=400)
try:
assert data, "Empty request body"
assert isinstance(data, dict), "Request body ain't a proper dict"
assert "ports" in data, "No 'ports' provided"
assert isinstance(data["ports"], list), "'ports' ain't a list"
assert len(data["ports"]) > 0, "'ports' list is empty"
assert len(data["ports"]) < 30, "That's too many ports to check"
assert len(data["ports"]) == len(set(data["ports"])), "'ports' list should contain unique elements"
def is_port_number(p):
return isinstance(p, int) and p > 0 and p < 65535
assert all(is_port_number(p) for p in data["ports"]), "'ports' should a list of valid port numbers"
except AssertionError as e:
logger.info(f"Invalid request: {e} ... Original request body was: {request.body}")
return json_response({
"error": {
"code": "error_bad_json_data",
"content": f"Invalid request: {e} ... Original request body was: {request.body}"
}
}, status=400)
# ############################################# #
# Run the actual check #
# ############################################# #
result = {}
for port in data["ports"]:
result[int(port)] = await check_port_is_open(ip, port)
return json_response({"ports": result})
async def check_port_is_open(ip, port):
if ":" in ip:
sock = socket.socket(socket.AF_INET6, socket.SOCK_STREAM)
else:
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.settimeout(2)
result = sock.connect_ex((ip, port))
sock.close()
return result == 0
# ########################################################################### #
# SMTP check #
# ########################################################################### #
@app.route("/check-smtp/", methods=["POST"])
async def check_smtp(request):
"""
This function received an HTTP request from a YunoHost instance while this
server is hosted on our infrastructure. The request is expected to be a
POST request with an empty body
The general workflow is the following:
- grab the ip from the request
- check for ip based rate limit (see RATE_LIMIT_SECONDS value)
- open a socket on port 25
- the server is supposed to say '200 domain.tld Service ready'
- we return the domain.tld found
"""
# this is supposed to be a fast operation if run often enough
now = time.time()
clear_rate_limit_db(now)
# ############################################# #
# Validate request and extract the parameters #
# ############################################# #
ip = request.headers["x-forwarded-for"].split(",")[0]
check_rate_limit_ip = check_rate_limit(ip, now)
if check_rate_limit_ip:
return check_rate_limit_ip
if ":" in ip:
sock = socket.socket(socket.AF_INET6, socket.SOCK_STREAM)
else:
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.settimeout(2)
result = sock.connect_ex((ip, 25))
if result != 0:
return json_response({
'status': "error_smtp_unreachable",
'content': "Could not open a connection on port 25, probably because of a firewall or port forwarding issue"
})
try:
recv = sock.recv(1024).decode('utf-8')
assert recv[:3] == "220"
helo_domain = recv.split()[1].strip()
except:
return json_response({
'status': "error_smtp_bad_answer",
'content': "SMTP server did not reply with '220 domain.tld' after opening socket ... Maybe another machine answered."
})
finally:
sock.close()
return json_response({'status': 'ok', 'helo': helo_domain})
@app.route("/")
async def main(request):
return html("You aren't really supposed to use this website using your browser.<br><br>It's a small server with an API to check if a services running on YunoHost instance can be reached from 'the global internet'.")
if __name__ == "__main__":
app.run(host="0.0.0.0", port=7000)

11
yunodiagnoser.service Normal file
View file

@ -0,0 +1,11 @@
[Unit]
Description=A server providing features for remote-diagnosis for Yunohost servers
[Service]
Type=simple
WorkingDirectory={{ WORKING_DIR }}
ExecStart={{ WORKING_DIR }}/venv/bin/python3.6 yunodiagnoser.py &> server.log
ExecStop=/bin/kill `/bin/ps aux | /bin/grep yunodiagnoser.py | /bin/grep -v grep | /usr/bin/awk '{ print $2 }'`
[Install]
WantedBy=multi-user.target