Redirect + datagroup

I couldn't say exactly why but this won't work, you're doing too much at once. This should;

 when HTTP_REQUEST { 
  Compare the request URI with the strings in the uri-dg data group 
  and populate the variable: redirect with the associated value if  
  there is a match 
  set redirect [class match -value [HTTP::uri] equals dg_redirect] 
     
   As long as our variable: redirect isn't empty, perform the redirect 
   if { $redirect ne "" } { 
    HTTP::redirect "http://[HTTP::host]$redirect" 
   }  
  
   If there wasn't a match and variable: redirect is empty, do something 
   else instead 
   else { 
    log local0. "No redirect occurred for: [HTTP::uri]"     
   } 
 } 

Keep in mind that data groups become temporarily unavailable when you save them. You'll want to build a catch into your code to make sure that the data group exists to avoid an error.

 

There are two conditions you may encounter for a brief period of time when you save it:

 

1. The data group doesn't exist yet as it's being recreated.
2. Not all data has been written to the group yet so some redirects will not work temporarily.

This happens for only a brief moment (the duration depends on the number of entries in the group), but it's something to keep in mind.

 

Hey Arie. I wasn't aware of that, I hate to question you but are you sure that's still the case?

Updated iRule (untested) incorporating the check;

when HTTP_REQUEST {
 Compare the request URI with the strings in the uri-dg data group
 and populate the variable: redirect with the associated value if
 there is a match
 if { class exists dg_redirect } {
  set redirect [class match -value [HTTP::uri] equals dg_redirect]
   As long as our variable: redirect isn't empty, perform the redirect
   if { $redirect ne "" } {
    HTTP::redirect "http://[HTTP::host]$redirect" 
   }  
  
   If there wasn't a match and variable: redirect is empty, do something 
   else instead 
   else { 
    log local0. "No redirect occurred for: [HTTP::uri]"
  }
 }
}

i understand these are discussion topics Arie has worked on.

 

 

Datagroup access during a datagroup update

 

https://devcentral.f5.com/community/group/aft/1172712/asg/50

 

 

Data Group (Class) unavailable during updates, causing errors

 

https://devcentral.f5.com/community/group/asg/50/afs/desc/aft/2160953/showtab/groupforums

Thanks Nitass.

Based on what I read, should my class check look like this instead;

if { [class exists dg_redirect] } {

Unfortunately, this is still an issue. Please do open cases with F5 Support and request to have your case attached to BZ381100 to raise the visibility of this bug.

 

 

Aaron

Shame, thanks for the confirmation Aaron.

Based on what I read, should my class check look like this instead; i never tested but if i do not read bug detail wrongly, it does not say we delete data group but just entry in data group. anyway, Aaron and Arie may be able to provide more information because they have worked on it.

 

 

Bug 381100 - Data group / class updates are not processed atomically by tmm

When I tested the datagroup always existed, but the elements were removed individually and then added back individually. So you'll have an incomplete data group for the change period but the object itself should always exist.

 

 

If it's critical that the datagroup contents always be whole during production hours, you'd want to limit changes to the datagroup to a maintenance window.

 

 

Aaron

Thanks gents.