[prev in list] [next in list] [prev in thread] [next in thread] 

List:       exim-users
Subject:    [exim] behavior of $address_data in routers
From:       Brian Blood <brian () ecmsquared ! com>
Date:       2008-10-24 2:35:41
Message-ID: 527229CD-DEC7-45FC-BB42-F148C6119805 () ecmsquared ! com
[Download RAW message or body]

Took a bit of debugging and strange router construction (can we please  
get logwrite to somehow be called inside a router), but I wanted to  
document what I found in relation to what happens to $address_data  
during routing, redirects and child routing.

I had very specific use case and the documentation explained what the  
pieces did, but not when I put them all together in a very specific way.

----

In our system a user account can be a mailbox account, a forwarding (1  
or more addresses) account or both.

In our router config at the point where we would redirect to any  
forwarding addresses, we test whether the on_forward bit is on AND  
then also test whether the on_fwd_keep bit is on and that test result  
is expanded for the unseen option for that router.

FYI This is PSEUDO router code.
For brevity's sake where you see: $address_data->field, in the actual  
config it looks something like this:
   ${extract{field}{$address_data}{$value}{0}}

and $is_local_address is something like:
   >{${extract{user_id}{$address_data}{$value}{0}}}{0}


forward:
  no_verify
  condition = $is_local_address AND ($address_data->on_forward > 0)
  driver = redirect
  check_ancestor
  qualify_preserve_domain
  data  = ${lookup mysql{SELECT forward FROM users \
		WHERE user_id = $rcpt_db_userid}{$value}{}}
  unseen = ($address_data->on_fwd_keep > 0)

local_user:
  no_verify
  condition = $is_local_address AND ($address_data->smtpdir != "")
  driver = accept
  retry_use_local_part
  transport = virtual_delivery
  address_data = $address_data
# pass on the entire record to the transport. let it extract
#   smtpdir and quota as necessary


mx_deliver:
  condition = !$is_local_address
  driver = dnslookup
  transport = remote_smtp



e.g. AddrA -> AddrB,AddrC (on_forward and on_fwd_keep are both set to  
ONE)

When exim is running an address through the routers and hits a  
redirect router that accepts the address, as documented, it starts a  
new routing chain going back to the top of the routers and starts  
testing each address returned by the "data" option of that original  
redirect router.

If you've set the $address_data variable at any point then at the  
beginning of each child router run, $address_data is populated with  
the value it had from the parent at the point of the redirect. In my  
setup, I'm doing a db lookup at the very first non-verifying router  
and setting $address_data to those results like so:

lookup_user_data:
  no_verify
  driver = redirect
  address_data = ${lookup mysql{SELECT  
user_id,on_forward,user_quota,on_fwd_keep,smtpdir \
         FROM users WHERE username=$local_part AND domain=$domain }}
  data =

...at this point $address_data has either the db field values for that  
users record or empty if the address in question is not a local domain

  What I really wanted to know was if I had set $address_data on each  
child routing run and unseen was set on the parent address, what was  
in $address_data after the child(ren) routing chains were completed  
and exim came back from the call chain to continue routing the parent  
address for addresses where unseen was set in the positive.

Empirical results:
  -- $address_data gets set at beginning of each child run to the  
parent value, regardless of what it is set to during a child run.

  -- $address_data retains it's original value for the parent when  
returning from child routing runs to finish the original routing  
process.

This makes my setup work very well, as looking up and setting  
$address_data at the very first router gives and maintains the data I  
went to the db for and I can rely on that data being there at the  
appropriate time.


This is a very specific instance, but I thought seeing it all laid out  
beginning to end might help someone later on even if you aren't doing  
this exact same thing.


Brian



-- 
## List details at http://lists.exim.org/mailman/listinfo/exim-users 
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/
[prev in list] [next in list] [prev in thread] [next in thread]