Haproxy route and rewrite based on URI path

Question

I am trying to setup an Haproxy to load balance requests on a few backends identified by the uri path. For example:

https://www.example.com/v1/catalog/foo/bar

Should lead to the "catalog-v1" backends.

Thing is each app responds on a different path so I must not only identify the app but rewrite the URL path. E.g.

• https://www.example.com/v1/catalog/product
• https://www.example.com/v2-2/checkout/cart/123
• https://www.example.com/v3.1.2/checkout/cart

TO

• https://www.example.com/catalog-v1/product
• https://www.example.com/checkout-v2-2/cart/123
• https://www.example.com/checkout-v3.1.2/cart

I know I shouldn't use Haproxy for rewriting purposes but for now this is inevitable.

Tried the following regex which worked on regex101:

([a-z.]*)\/([a-z0-9\-\.]*)\/([a-z\-]*)\/(.*)

Substitution:

\1/\3-\2/\4

And finally here's the haproxy.config:

global
    daemon
    user root
    group root
    maxconn 256000
    log     127.0.0.1 local0
    log     127.0.0.1 local1 notice
    stats   socket /run/haproxy/stats.sock mode 777 level admin
defaults
    log      global
    option   dontlognull
    maxconn  4000
    retries  3
    timeout  connect 5s
    timeout  client  1m
    timeout  server  1m
    option   redispatch
    balance  roundrobin

listen stats :8088
    mode http
    stats enable
    stats uri /haproxy
    stats refresh 5s

backend catalog-v1
    mode http
    option httpchk GET /catalog-v1/ping
    http-check expect status 200
    reqrep ([a-z.]*)\/([a-z0-9\-\.]*)\/([a-z\-]*)\/(.*)   \1/\3-\2/\4
    server 127.0.0.1:8280_catalog-v1-node01 127.0.0.1:8280 check inter 2s rise 3 fall 2

backend checkout-v1
    mode http
    option httpchk GET /checkout-v1/ping
    http-check expect status 200
    reqrep ([a-z.]*)\/([a-z0-9\-\.]*)\/([a-z\-]*)\/(.*)   \1/\3-\2/\4
    server 127.0.0.1:8180_checkout-v1-node01 127.0.0.1:8180 check inter 2s rise 3 fall 2

frontend shared-frontend
    mode http
    bind localhost:80
    acl is-catalog-v1-path path_dir /v1/catalog
    acl is-checkout-v1-path path_dir /v1/checkout
    use_backend catalog-v1 if is-catalog-v1-path
    use_backend checkout-v1 if is-checkout-v1-path

Am I missing something?

I've been struggling with this for quite some time without success. The backends shows "UP" in Haproxy stats page but everytime I call the "non rewrited url" I get a 400 Bad Request error.

Thanks in advance for your help!

Answer 1

The answer above is a single fix that doesn't explain the root of the problem. The problem is that many assume we're parsing a complete URL or only the PATH component, when we're actually parsing/rewriting HTTP headers.

For example:

curl -iv https://stackoverflow.com/questions/24784517/haproxy-route-and-rewrite-based-on-uri-path
*   Trying 151.101.65.69...
> GET /questions/24784517/haproxy-route-and-rewrite-based-on-uri-path HTTP/1.1
> Host: stackoverflow.com
> User-Agent: curl/7.54.0
> Accept: */*

The goal is to rewrite GET /some_path HTTP/1.1 to GET /some_other_path HTTP/1.1

I want to clarify that the solution above works because it takes into account the HTTP verb in the match and replace. ^([^\ ]\*)\ (.*) to capture the verb and use it in the replacement pattern. \1 \2

Answer 2

If I understand you correctly, replace the example below:

reqrep ^([^\ ]\*)\ /([-.0-9A-Za-z]\*)/([a-zA-Z]\*)/(.\*)  \1\ /\3-\2/\4

http://cbonte.github.io/haproxy-dconv/configuration-1.5.html#reqrep

reqrep search string [{if | unless} cond]

You have no spaces.