I'm having some trouble with Exim that seems to be caused by the "unseen" directive in virtual_aliases_nostar director in the WHM Exim configuration, advanced editor.
Background:
I have a website on server2.example.com and a mail server on server1.example.com, and I want to use server1 as a smarthost for server2. So I added this to te "PREROUTERS" section in Exim, advanced editor, on server2 (adapted from Howto: redirect a copy of your emails to another server while saving it locally):
smarthost:
driver = manualroute
domains = !+local_domains
transport = remote_smtp
headers_add = X-Custom-Forwarded: true
condition = ${if match {$h_X-Custom-Forwarded:}{true}{false}{true}}
route_data = "server1.example.com"
That worked for all domains that existed on both servers.
However, I have some domains that have their websites on server2, and only their DNS on server1. To get the smarthost to work for those domains, I had to add the domain name to /etc/localdomains and create a corresponding /etc/valiases file.
After I did that, the mail sent from server2 was successfully delivered by server1, but the sender also got a failure notice.
I entered the address into the Mail troubleshooter on server1 and got something like this:
[email protected]
virtual_aliases_nostar via /etc/userdomains
[email protected]
[email protected]
virtual_aliases via virtual_aliases router forced address failure
In other words, the message was going first to virtual_aliases_nostar, then to virtual_aliases (apparently because of the "unseen" directive in virtual_aliases_nostar). It was delivered by virtual_aliases_nostar but then failed by virtual_aliases.
Solution:
It doesn't seem possible to edit virtual_aliases_nostar, so instead I added this to the "PREVALIASNOSTAR" section just above virtual_aliases_nostar:
virtual_aliases_nostar1:
driver = redirect
allow_defer
allow_fail
require_files = "+/etc/valiases/$domain"
address_data = ${lookup{[email protected]$domain}lsearch{/etc/valiases/$domain}}
data = $address_data
file_transport = address_file
group = "${lookup{$domain}lsearch* {/etc/userdomains}{$value}}"
user = "${lookup{$domain}lsearch* {/etc/userdomains}{$value}}"
pipe_transport = ${if forall{/bin/cagefs_enter:/usr/sbin/cagefsctl}{exists{$item}}{cagefs_virtual_address_pipe}{${if match{${extract{6}{:}{${lookup passwd{${lookup{$domain}lsearch*{/etc/userdomains}{$value}}}{$value}}}}}{\N(jail|no)shell\N}{jailed_virtual_address_pipe}{virtual_address_pipe}}}}
retry_use_local_part
Basically, that's an exact copy of virtual_aliases_nostar, minus the "unseen" directive. It seems to fix my specific problem with no immediately visible side effects.
Question:
Is this going to cause other problems? And is there a better solution?
FWIW, all of the addresses for these domains are forwarders in the /etc/valiases files; there are no actual mailboxes. Autoresponders still seem to work OK.
Background:
I have a website on server2.example.com and a mail server on server1.example.com, and I want to use server1 as a smarthost for server2. So I added this to te "PREROUTERS" section in Exim, advanced editor, on server2 (adapted from Howto: redirect a copy of your emails to another server while saving it locally):
smarthost:
driver = manualroute
domains = !+local_domains
transport = remote_smtp
headers_add = X-Custom-Forwarded: true
condition = ${if match {$h_X-Custom-Forwarded:}{true}{false}{true}}
route_data = "server1.example.com"
That worked for all domains that existed on both servers.
However, I have some domains that have their websites on server2, and only their DNS on server1. To get the smarthost to work for those domains, I had to add the domain name to /etc/localdomains and create a corresponding /etc/valiases file.
After I did that, the mail sent from server2 was successfully delivered by server1, but the sender also got a failure notice.
I entered the address into the Mail troubleshooter on server1 and got something like this:
[email protected]
virtual_aliases_nostar via /etc/userdomains
[email protected]
[email protected]
virtual_aliases via virtual_aliases router forced address failure
In other words, the message was going first to virtual_aliases_nostar, then to virtual_aliases (apparently because of the "unseen" directive in virtual_aliases_nostar). It was delivered by virtual_aliases_nostar but then failed by virtual_aliases.
Solution:
It doesn't seem possible to edit virtual_aliases_nostar, so instead I added this to the "PREVALIASNOSTAR" section just above virtual_aliases_nostar:
virtual_aliases_nostar1:
driver = redirect
allow_defer
allow_fail
require_files = "+/etc/valiases/$domain"
address_data = ${lookup{[email protected]$domain}lsearch{/etc/valiases/$domain}}
data = $address_data
file_transport = address_file
group = "${lookup{$domain}lsearch* {/etc/userdomains}{$value}}"
user = "${lookup{$domain}lsearch* {/etc/userdomains}{$value}}"
pipe_transport = ${if forall{/bin/cagefs_enter:/usr/sbin/cagefsctl}{exists{$item}}{cagefs_virtual_address_pipe}{${if match{${extract{6}{:}{${lookup passwd{${lookup{$domain}lsearch*{/etc/userdomains}{$value}}}{$value}}}}}{\N(jail|no)shell\N}{jailed_virtual_address_pipe}{virtual_address_pipe}}}}
retry_use_local_part
Basically, that's an exact copy of virtual_aliases_nostar, minus the "unseen" directive. It seems to fix my specific problem with no immediately visible side effects.
Question:
Is this going to cause other problems? And is there a better solution?
FWIW, all of the addresses for these domains are forwarders in the /etc/valiases files; there are no actual mailboxes. Autoresponders still seem to work OK.
