The suggested configurations for nginx found in the documentation and templates lead to vulnerabilities allowing host spoofing [1] and path traversal [2], as reported by Gixy [3]. This commit fixes those issues. [1] https://github.com/yandex/gixy/blob/master/docs/en/plugins/hostspoofing.md [2] https://github.com/yandex/gixy/blob/master/docs/en/plugins/aliastraversal.md [3] https://github.com/yandex/gixy
5.9 KiB
How to protect an instance
further reading
filtron.sh
nginx searx site
Contents
Searx depends on external search services. To avoid the abuse of these services it is advised to limit the number of requests processed by searx.
An application firewall, filtron solves exactly this problem. Filtron is just a middleware between your web server (nginx, apache, ...) and searx, we describe such infratructures in chapter: architecture
.
filtron & go
Filtron needs Go installed. If Go is preinstalled, filtron is simply installed by go get
package management (see filtron README). If you use filtron as middleware, a more isolated setup is recommended. To simplify such an installation and the maintenance of, use our script filtron.sh
.
Sample configuration of filtron
Tooling box
/etc/filtron/rules.json <utils/templates/etc/filtron/rules.json>
An example configuration can be find below. This configuration limits the access of:
- scripts or applications (roboagent limit)
- webcrawlers (botlimit)
- IPs which send too many requests (IP limit)
- too many json, csv, etc. requests (rss/json limit)
- the same UserAgent of if too many requests (useragent limit)
[
{
"name": "search request",
"filters": [
"Param:q",
"Path=^(/|/search)$"
],
"interval": "<time-interval-in-sec (int)>",
"limit": "<max-request-number-in-interval (int)>",
"subrules": [
{
"name": "missing Accept-Language",
"filters": ["!Header:Accept-Language"],
"limit": "<max-request-number-in-interval (int)>",
"stop": true,
"actions": [
{"name":"log"},
{"name": "block",
"params": {"message": "Rate limit exceeded"}}
]
},
{
"name": "suspiciously Connection=close header",
"filters": ["Header:Connection=close"],
"limit": "<max-request-number-in-interval (int)>",
"stop": true,
"actions": [
{"name":"log"},
{"name": "block",
"params": {"message": "Rate limit exceeded"}}
]
},
{
"name": "IP limit",
"interval": "<time-interval-in-sec (int)>",
"limit": "<max-request-number-in-interval (int)>",
"stop": true,
"aggregations": [
"Header:X-Forwarded-For"
],
"actions": [
{ "name": "log"},
{ "name": "block",
"params": {
"message": "Rate limit exceeded"
}
}
]
},
{
"name": "rss/json limit",
"filters": [
"Param:format=(csv|json|rss)"
],
"interval": "<time-interval-in-sec (int)>",
"limit": "<max-request-number-in-interval (int)>",
"stop": true,
"actions": [
{ "name": "log"},
{ "name": "block",
"params": {
"message": "Rate limit exceeded"
}
}
]
},
{
"name": "useragent limit",
"interval": "<time-interval-in-sec (int)>",
"limit": "<max-request-number-in-interval (int)>",
"aggregations": [
"Header:User-Agent"
],
"actions": [
{ "name": "log"},
{ "name": "block",
"params": {
"message": "Rate limit exceeded"
}
}
]
}
]
}
]
Route request through filtron
further reading
filtron.sh overview
installation nginx
installation apache
Filtron can be started using the following command:
filtron -rules rules.json $
It listens on 127.0.0.1:4004
and forwards filtered requests to 127.0.0.1:8888
by default.
Use it along with nginx
with the following example configuration.
# https://example.org/searx
location /searx {
proxy_pass http://127.0.0.1:4004/;
proxy_set_header Host $host;
proxy_set_header Connection $http_connection;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Scheme $scheme;
proxy_set_header X-Script-Name /searx;
}
location /searx/static {
/usr/local/searx/searx-src/searx/static;
}
Requests are coming from port 4004 going through filtron and then forwarded to port 8888 where a searx is being run. For a complete setup see: nginx
searx site
.