Clock time

Well, first, I would not recommend to heavily using the clock command until 9.2 as it does have a potential performance impact (however, it may not effect you enough to be concerned about it).

The main command to use is [clock seconds] which will return you the number of seconds since epoch. You can use this to easily compute the number of seconds that a pool member has been down. To then format this into something more human readable you would use the [clock format] command. Eg:

when RULE_INIT {
   array set ::down_nodes { }
}
when LB_FAILED {
   set cur_time [clock seconds]
   set server [LB::server addr]
    Has this member been down for more than one request
   if { [info exists ::down_nodes($server)] } {
      set delta [expr $cur_time - $::down_nodes($server)]
       Start logging after the member has been down for more than 60 secs
      if { $delta > 60 } {
         log local0. "Pool member [LB::server] down [clock format $delta -format {%H:%M:%S}]"
      }
   }
   set ::down_nodes($server) $cur_time
}
when LB_SELECTED {
    Clear out a down node that is now up
   if { [info exists ::down_nodes([LB::server addr])] } {
      unset ::down_nodes([LB::server addr])
   }
}