Forum Discussion

msaud

Cirrus

Jun 19, 2023

F5 iRUule not working

This is API Application rather than web browser.

Source URL:

https://mule-dev-openai-lb.ucsf.edu/tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01

Destination URL:

https://tccog2.openai.azure.com/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01

tccog2 – azure openai instance name

irule Requirements:

1) destination host translation (prepend openai.azure.com with azure openai instance name) Immediate resource after the host name in the source url is extracted and prepended in the dest host name. As shown in the example above tccog2 is extracted from source url and prepended in the dest host name tccog2.openai.azure.com.

2) uri path translation (remove the azure openai instance name from uri path) Rest of the uri path (/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01) after the first resource from the host name is extracted and send us the destination uri path after the host name.

Following irule is not working to satisfy the above requirements:

when HTTP_REQUEST {

if { [HTTP::host] equals "mule-dev-openai-lb.ucsf.edu"} {

if { [HTTP::path] equals "/tccog2" } {

set newUri [string range [HTTP::uri] [string first / [HTTP::uri] 1] end]

set HTTP::uri ${newUri}

HTTP::redirect "https://tccog2.openai.azure.com[HTTP::uri]"

}

ScottE
Jun 19, 2023
Hello msaud

Please remove the "set" in the "set HTTP::uri ${newUri}" command so it reads "HTTP::uri ${newUri}". I usually don't have the brackets but I doubt that makes a difference.

Scott

AaronJB

Jun 19, 2023

[edit] Ignore this, ScottE had the proper advice below 🙂

What's not working for you exactly - what behaviour are you seeing?

I did a quick test in the lab and I see the behaviour you'd expect (a 302 redirect) - obviously I've set the IP for mule-dev-openai-lb.ucsf.edu to the VIP in my lab here:

$ curl -kv 'http://mule-dev-openai-lb.ucsf.edu/tccog2/openai/deployments/tccognitivedeployment/completi
ons?api-version=2022-12-01'
*   Trying 10.10.16.10:80...
* Connected to mule-dev-openai-lb.ucsf.edu (10.10.16.10) port 80 (#0)
> GET /tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01 HTTP/1.1
> Host: mule-dev-openai-lb.ucsf.edu
> User-Agent: curl/7.81.0
> Accept: */*
>
* Mark bundle as not supporting multiuse
* HTTP 1.0, assume close after body
< HTTP/1.0 302 Moved Temporarily
< Location: https://tccog2.openai.azure.com/tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01
< Server: BigIP
* HTTP/1.0 connection set to keep alive!
< Connection: Keep-Alive
< Content-Length: 0
<
* Connection #0 to host mule-dev-openai-lb.ucsf.edu left intact

ScottE
MVP
Jun 19, 2023
Hello msaud

Please remove the "set" in the "set HTTP::uri ${newUri}" command so it reads "HTTP::uri ${newUri}". I usually don't have the brackets but I doubt that makes a difference.

Scott
- AaronJB
  SIRT
  Jun 19, 2023
  Whoops, I missed that my test hadn't rewritten the URI - I can confirm as well, remove the 'set' and it works for me:
  
  $ curl -kv 'http://mule-dev-openai-lb.ucsf.edu/tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01' * Trying 10.10.16.10:80... * Connected to mule-dev-openai-lb.ucsf.edu (10.10.16.10) port 80 (#0) > GET /tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01 HTTP/1.1 > Host: mule-dev-openai-lb.ucsf.edu > User-Agent: curl/7.81.0 > Accept: */* > * Mark bundle as not supporting multiuse * HTTP 1.0, assume close after body < HTTP/1.0 302 Moved Temporarily < Location: https://tccog2.openai.azure.com/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01 < Server: BigIP * HTTP/1.0 connection set to keep alive! < Connection: Keep-Alive < Content-Length: 0 < * Connection #0 to host mule-dev-openai-lb.ucsf.edu left intact
  - msaud
    Cirrus
    Jun 20, 2023
    Hi Aaron,
    Look like iRule is working but we are getting below errors when trying to connect VIP-
    "error":{"code":"404","message": "Resource not found"}}
    
    Is there other things that we need to check?
    10:48:45.455 06/19/2023 Worker-0 [MuleRuntime].uber.05: [stagegw-its-azure-openai-proxy].proxy.CPU_LITE @7e88db99 DEBUG
    event:898e3e00-0ec9-11ee-a321-02a4b434dd63 REQUESTER POST /tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01 HTTP/1.1 x-ssl-client-verify: NONE x-real-ip: 64.54.198.22 user-agent: PostmanRuntime/7.32.2 accept: */* cache-control: no-cache postman-token: bb49fc79-8441-4426-b4ea-c7b2de2835c7 accept-encoding: gzip, deflate, br authorization: Basic MDYxNzgyMjM5YjA3NDRlNDg3Y2U0ZWI0MTEzMmE0NTA6RDJhMTQ4MDQ1MEM0NDQzQTg0MzE4MTUyNGMwZUMxNjA= x-forwarded-for: 64.54.198.22, 10.225.46.165 x-forwarded-host: /10.225.46.190:60294 api-key: 03308343f0004146bc35159c0b624c3c x-anypnt-app-worker: A802BD108B50D526A4D3F6A5B49CDA97 x-correlation-id: 898e3e00-0ec9-11ee-a321-02a4b434dd63 Host: mule-stage-openai-lb.ucsf.edu:443 Connection: keep-alive Content-Type: application/json Content-Length: 53 { "prompt": "Once upon a time", "max_tokens": 5 }
    10:48:45.564 06/19/2023 Worker-0 [stagegw-its-azure-openai-proxy].http.requester.http-request-config.01 SelectorRunner DEBUG
    event:898e3e00-0ec9-11ee-a321-02a4b434dd63 REQUESTER HTTP/1.1 404 Resource Not Found Date: Mon, 19 Jun 2023 17:48:45 GMT Content-Type: application/json Content-Length: 56 Connection: keep-alive apim-request-id: a7f2e0b9-fd83-4d2e-a36b-71729cc18a4f Strict-Transport-Security: max-age=31536000; includeSubDomains; preload x-content-type-options: nosniff {"error":{"code":"404","message": "Resource not found"}}
ScottE
MVP
Jun 19, 2023
Hello msaud

For source link: https://mule-dev-openai-lb.ucsf.edu/tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01

HTTP::path (path and filename in request) = /tccog2/openai/deployments/tccognitivedeployment/completions

so the following test

if { [HTTP::path] equals "/tccog2" } {

will not be true. Hopefully you can change that "equals" to "starts_with" and then you should be able to get to your setting of the newUri. If that is not behaving then you can log the current uri and the newUri to see ensure it is behaving as expected.

Scott
- msaud
  Cirrus
  Jun 19, 2023
  Hi Scotte,
  Thank you for your support on this. I did try below as well but still not working.
  when HTTP_REQUEST {
  if { [HTTP::host] equals "mule-dev-openai-lb.ucsf.edu"} {
  if { [HTTP::path] starts_with "/tccog2" } {
  set newUri [string range [HTTP::uri] [string first / [HTTP::uri] 1] end]
  set HTTP::uri ${newUri}
  HTTP::redirect "https://tccog2.openai.azure.com[HTTP::uri]"
  }
  }
  }
  - AaronJB
    SIRT
    Jun 19, 2023
    [edit] Ignore this, ScottE had the proper advice below 🙂
    
    What's not working for you exactly - what behaviour are you seeing?
    
    I did a quick test in the lab and I see the behaviour you'd expect (a 302 redirect) - obviously I've set the IP for mule-dev-openai-lb.ucsf.edu to the VIP in my lab here:
    
    $ curl -kv 'http://mule-dev-openai-lb.ucsf.edu/tccog2/openai/deployments/tccognitivedeployment/completi ons?api-version=2022-12-01' * Trying 10.10.16.10:80... * Connected to mule-dev-openai-lb.ucsf.edu (10.10.16.10) port 80 (#0) > GET /tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01 HTTP/1.1 > Host: mule-dev-openai-lb.ucsf.edu > User-Agent: curl/7.81.0 > Accept: */* > * Mark bundle as not supporting multiuse * HTTP 1.0, assume close after body < HTTP/1.0 302 Moved Temporarily < Location: https://tccog2.openai.azure.com/tccog2/openai/deployments/tccognitivedeployment/completions?api-version=2022-12-01 < Server: BigIP * HTTP/1.0 connection set to keep alive! < Connection: Keep-Alive < Content-Length: 0 < * Connection #0 to host mule-dev-openai-lb.ucsf.edu left intact