matchclass versus class in v10 for address comparisons?

I was reading in the wiki under matchclass that matchclass is deprecated in version 10. Also, it said that if you do something like this: [matchclass [HTTP::uri] starts_with $::my_data_class] that $::my_data_class name will now return the name of the data class instead of a list.

 

 

I understand that using class is better but what about compatibility for upgrading. If what I read is true, I have to rewrite a large number of iRules before I come up on version 10. I am running version 9.6 now.

 

 

I would appreciate it if anyone could clarify.

 

 

Regards,

 

 

Tom Schaefer

Hi Tom,

The minimal change you have to make is to remove the $:: prefix from the datagroup names in the iRules. If you do this you can continue to use findclass and/or matchclass in v10 or v11 as you did in v9.

To gain performance you should ideally change matchclass and findclass to the corresponding class sub-commands. You can also take advantage of the new name=value pairs available in 10.1+ using the class command with the -value option. This last option would require changing your string datagroups from a "key value" space separated format to the new name=value format:

class string_name_value_class {
   {
      "key1" { "value1" }
      "key2" { "value2" }
   }
}
class address_name_value_class {
   {
      host 1.1.1.1 { "value1" }
      network 1.1.2.0/24 { "value2" }
   }
}
 Sample iRule which gets the value for a string datagroup entry and an address datagroup entry
when RULE_INIT {
   set value [class match -value "key1" equals string_name_value_class]
   log local0. "Looked up key1 and found $value"
   set value [class match -value 1.1.2.25 equals address_name_value_class]
   log local0. "Looked up 1.1.2.25 and found $value"
}

Log output:

Looked up key1 and found value1

Looked up 1.1.2.25 and found value2

Aaron

Thanks. I do things now like have multiple data elements related to the row separated by a vertical bar. I then use getfield to break it out. So, if I do a findclass, I then parse the row returned (all after the key in the case of findclass). Is there asimilar concept in that I can have multiple name and valus pairs on a row or should I get away from the row mentality?

 

 

Thanks,

 

 

Tom

Sure. You can just move that to the value portion:

class string_name_value_class {
   {
      "key1" { "value1a|value1b|value1c" }
      "key2" { "value2a|value2b|value2c" }
   }
}

Aaron

 

 

Hi Tom,

 

 

The closing Bracket is missing in your Example :-)

 

class string_name_value_class {

 

{

 

"key1" { "value1" }

 

"key2" { "value2" }

 

}

 

}

 

 

class address_name_value_class {

 

{

 

host 1.1.1.1 { "value1" }

 

network 1.1.2.0/24 { "value2" }

 

}

 

}

 

 

Sample iRule which gets the value for a string datagroup entry and an address datagroup entry

 

when RULE_INIT {

 

 

set value [class match -value "key1" equals string_name_value_class]

 

log local0. "Looked up key1 and found $value"

 

 

set value [class match -value 1.1.2.25 equals address_name_value_class]

 

log local0. "Looked up 1.1.2.25 and found $value"

 

}

 

 

 

 

Regards

 

 

wiesmann

 

 

Hi Juerg,

 

 

That was my bad copy/paste. Thanks for pointing it out. I edited my post.

 

 

Aaron