Redirect to another URL from inside HTTP::REQUEST after timeout detected

Can you post the iRule so that we can see it?

can you try this?

 

 

...

 

log local0.debug "2. Redirecting to $theUrl"

 

both redirect and respond give "Illegal argument. Can't execute in the current context. (line 1)", TCP::respond has no effect

 

HTTP::redirect $theUrl

 

HTTP::respond 302 Location "$theUrl"

 

TCP::respond "HTTP/1.1 302 Found\r\nLocation: $theUrl\r\nConnection: close\r\nContent-Length: 0\r\n\r\n"

 

}\

 

...

 

 

this is mine.

 

 

[root@ve1023:Active] config curl -I http://172.28.19.79

 

HTTP/1.1 302 Found

 

Location:

 

Connection: close

 

Content-Length: 0

 

[root@ve1023:Active] config

 

Nov 30 11:47:19 local/tmm debug tmm[23027]: Rule myrule HTTP_REQUEST: F5 Host: 172.28.19.79, /, 80 ... Server: :

 

Nov 30 11:47:19 local/tmm debug tmm[23027]: Rule myrule LB_SELECTED: In lb_selected, server is 200.200.200.101:80 ...

 

Nov 30 11:47:19 local/tmm info tmm[23027]: Rule myrule HTTP_REQUEST: *** Timeout after 3 ms ***

 

Nov 30 11:47:19 local/tmm debug tmm[23027]: Rule myrule HTTP_REQUEST: 1. Taking node 200.200.200.101:80 out of the mix ...

 

Nov 30 11:47:19 local/tmm err tmm[23027]: 01010028:3: No members available for pool foo

 

Nov 30 11:47:19 local/tmm debug tmm[23027]: Rule myrule HTTP_REQUEST: 2. Redirecting to

 

Nov 30 11:47:19 local/ve1023 err mcpd[21973]: 01070809:3: iRule has marked pool member 200.200.200.101:80 down, no monitor to mark up.

 

Nov 30 11:47:19 local/ve1023 notice mcpd[21973]: 01070638:5: Pool member 200.200.200.101:80 monitor status iRule down.

 

Hi Roman,

 

 

I have been looking at your iRule. The first thing that I am noticing is that you are setting variables in the RULE_INIT event that are being reset in the CLIENT_ACCEPTED Event and then again in the HTTP_REQUEST Event.

 

 

The RULE_INIT is triggered when an iRule is added or is modified.

 

The CLIENT_ACCEPTED is triggered for each client when they connect to the Virtual Server hosting the iRule the first time.

 

The HTTP_REQUEST for each individual request.

 

 

So you are doing some work twice for nothing when you could just do them in the HTTP_REQUEST event (since you are resetting them for each individual request anyway).

 

 

These are the ones I am referring to:

 

set theUrl ""

 

set retry 0

 

set server_addr ""

 

set server_port ""

 

set host_name ""

 

set host_uri ""

 

set host_port ""

 

set host_port_str ""

 

set host_protocol ""

 

set postBackUrl ""

 

 

As far as your error, can reproduce the error, but I am wondering if the illegal argument error that you are getting is due to something missing.

 

 

What does the logging statement show in your LTM Log:

 

log local0.debug "2. Redirecting to $theUrl"

 

 

Does it appear to be the correct URL that it should be?

 

Hi nitass,

 

 

Thanks for your reply. Yes, I've tried the "&", and LTM has choked on the syntax:

 

Rule Smart_LB_Rule : 1. Taking node 10.2.178.121:8001 out of the mix ...

 

TCL error: Smart_LB_Rule - invalid command name "host=10.2.178.121&" while executing "host=$server_addr&"

 

 

Funny thing is that this scenario works fine inside LB_FAILED (when server is down), and I do get redirected to the controller url. The same goes for HTTP_RESPONSE when I detect http code > 500. I have a feeling it's the "after" command pecularities which prevent me from doing redirect.

 

 

Thank you,

 

Roman

 

 

PS: You must have edited your post while I've been trying your suggestion :)

Yes, I've tried the "&", and LTM has choked on the syntax:sorry, have you tried TCP::respond i posted? it seems working in my lab unit.

 

 

TCP::respond "HTTP/1.1 302 Found\r\nLocation: $theUrl\r\nConnection: close\r\nContent-Length: 0\r\n\r\n"

 

 

PS: You must have edited your post while I've been trying your suggestion :)yes, it seems devcentral does not like triangle bracket. so, i had to edit my post. :-)

Hi Michael,

 

 

Thank you very much for looking into my messy rule. Yes, you're abs right, and I do need a lot of cleaning to do. I will as soon as I get it working, I promise.

 

 

As far as the url, this is what I see in the logs (masking sensitive data):

 

Rule Smart_LB_Rule : In lb_selected, server is server.ip.address:8001 ...

 

Rule Smart_LB_Rule : *** Timeout after 15000 ms ***

 

Rule Smart_LB_Rule : 1. Taking node server.ip.address:8001 out of the mix ...

 

Rule Smart_LB_Rule : 2. Redirecting to http://domain_name.company_name.com/controller/?source=ltm&action=restart&host=server.ip.address&service=/bad/&port=8001&name=&cause=no_response_from_server_after_15000_ms&postBack=http://domain_name.company_name.com/bad/

 

TCL error: Smart_LB_Rule - Illegal argument. Can't execute in the current context. (line 1) invoked from within "HTTP::redirect $theUrl "

 

 

The url is valid, and if I put it in the browser, it works just fine. It also works (the HTTP::redirect $theUrl command) from inside both LB_FAILED and HTTP_RESPONSE events (as I've described to nitass). I have a sneaky suspicion it's the "after" timer that is giving me a trouble.

 

 

Thank you,

 

Roman

Hi nitass,

 

 

Sorry, I didn't see TCP::respond in your post. I've tested my rule with it a dozen times, and it seems to be working fine! I'm pretty sure I've tried the TCP::respond beforehand, but I guess I didn't have the complete syntax

 

 

Thank you very much for your help!

 

Roman

The solution is there, but did anyone ever determine why the HTTP::redirect is being executed within the CLIENT_ACCEPTED context, as the error message states? Why isn't it being executed only within the HTTP_RESPONSE or HTTP_REQUEST contexts?

 

 

What is it about using "after" and a timeout period that seems to cause the code to be run in the CLIENT_ACCEPTED context?

 

 

I have a similar rule that uses an after that periodically triggers a similar error (only difference is that I'm trying to use HTTP::respond). I put the command in the HTTP_REQUEST context, yet it's being triggered in the CLIENT_ACCEPTED context.

not sure if i catch you correctly. anyway, i am also wondering what context after's script is running is.

 

 

for example, when using HTTP::redirect instead of TCP::respond in the following example, i got context error.

 

 

send response from BigIP if server does not respond within a specified amount of time

 

https://devcentral.f5.com/wiki/iRules.after.ashx

 

 

/var/log/ltm

 

Aug 10 15:30:03 local/tmm info tmm[4860]: Rule myrule : Timeout 1000 milliseconds elapsed without server response. 1344583803

 

Aug 10 15:30:03 local/tmm err tmm[4860]: 01220001:3: TCL error: myrule - Illegal argument. Can't execute in the current context. (line 1) invoked from within "HTTP::redirect "http://www.google.com/""

 

 

i think i will open a case to verify unless someone here can give me a shot. :-)