Forum Discussion
Chadri_103513
Nimbostratus
Nimbostratusot Payload data npassing with POST inspection
Hi,
I need some help with an iRule that is designed to inspect both GET and POST data, compare it against a data-group of accepted commands, and then redirect it to the appropriate pool. The GET portion of this iRule is working, however, I am having issues with the POST portion. While debugging the iRule, I see that the POST portion is inspecting the traffic, matching the data-group, and rewriting the header but the payload doesn't appear to be making it back to the server. How do I make sure that all payload data is making it back to the server?
when HTTP_REQUEST {
set exampleHost "example.staging.local"
set examplePath "/exampleproxy/exampleproxy.aspx"
set examplePool "example.staging-80"
HTTP::header insert "OLD_HOST" [HTTP::host]
if { [HTTP::version] eq "1.1" } {
if { [HTTP::header is_keepalive] } {
HTTP::header replace "Connection" "Keep-Alive"
}
HTTP::version "1.0"
}
if { [HTTP::host] eq $exampleHost } {
pool $examplePool
return
}
elseif { [string tolower [HTTP::path]] eq "$examplePath"} {
HTTP::header replace "Host" $exampleHost
pool $examplePool
return
}
if { [HTTP::method] eq "GET"} {
set cmd [ URI::query [string tolower [HTTP::uri]] command ];
if { [matchclass $::example_staging equals $cmd] } {
HTTP::header replace "Host" $exampleHost
HTTP::path $examplePath
pool $examplePool
return
}
}
elseif { [HTTP::method] eq "POST" } {
if { [HTTP::header exists "Content-Length"] } {
HTTP::collect [HTTP::header Content-Length]
} else {
HTTP::collect 256
}
}
}
when HTTP_REQUEST_DATA {
set cmd [string tolower [getfield [lsearch -inline [split [HTTP::payload] "&"] command=*] = 2]]
if {[matchclass $::example_staging equals $cmd]} {
HTTP::header replace "Host" $exampleHost
HTTP::path $examplePath
set payload [HTTP::payload]
pool $examplePool
return
}
}
Any help would be greatly appreciated.
Thanks.
3 Replies
Nat_ThirasuttakornEmployee
maybe HTTP::release can help
(put it before return inside HTTP_REQUEST_DATA. actually, you don't need return command there)
Nat
hoolioCirrostratus
Hi,
I'd be suspicious about the default of collecting 256 bytes for POST requests without a Content-Length header. Can you add logging to that portion of the rule:... elseif { [HTTP::method] eq "POST" } { log local0. "[IP::client_addr]:[TCP::client_port]: POST request to [HTTP::host][HTTP::uri]\ with Content-Length: [HTTP::header Content-Length]" if { [HTTP::header exists "Content-Length"] } { HTTP::collect [HTTP::header Content-Length] log local0. "[IP::client_addr]:[TCP::client_port]: Collecting [HTTP::header Content-Length] bytes" } else { HTTP::collect 256 log local0. "[IP::client_addr]:[TCP::client_port]: No Content-Length, collecting 256 bytes: Req: [HTTP::request]" } } when HTTP_REQUEST_DATA { log local0. "[IP::client_addr]:[TCP::client_port]: Collected [HTTP::payload length] bytes: [HTTP::payload]" Parse the command from the request payload set cmd [string tolower [getfield [lsearch -inline [split [HTTP::payload] "&"] command=*] = 2]] set cmd [string tolower [URI::query "?[HTTP::payload]" command]] if {[matchclass $::example_staging equals $cmd]} { log local0. "[IP::client_addr]:[TCP::client_port]: Matched $cmd, replacing host header, path and setting pool." HTTP::header replace "Host" $exampleHost HTTP::path $examplePath pool $examplePool } }
Assuming you're parsing parameter names and values that are URL encoded in the format param1=value1¶m2=value2, you could replace this line:
set cmd [string tolower [getfield [lsearch -inline [split [HTTP::payload] "&"] command=*] = 2]]
with this line:
set cmd [string tolower [URI::query "?[HTTP::payload]" command]]
I would guess the inbuilt processing of the parameters would be faster than the list commands.
Aaron- hoolioHi Nat,
The wiki page for HTTP::release says:
http://devcentral.f5.com/wiki/default.aspx/iRules/http__release
Releases the data collected via HTTP::collect. Unless a subsequent HTTP::collect command was issued, there is no need to use the HTTP::release command inside of the HTTP_REQUEST_DATA and HTTP_RESPONSE_DATA events, since (in these cases) the data is implicitly released.
Do you think it still might be required? I was thinking the issue might be an atypical client sending a chunked request. In that case, the request wouldn't have a Content-Length header and could be hitting the HTTP::collect 256 statement.
This seems like an odd problem, as the payload isn't being modified in any way...
Thanks,
Aaron
