Re: Patch for new feature...

2 msg	Envelope sender problem
10 msg	activating mailman
5 msg	bounces and sender_dependent_relayhost_maps
4 msg	Clustering Postfix
3 msg	Splitting delivery of one domain?
4 msg	Sending 5.000 mails to an 'alias' user
2 msg	move spam to IMAP spam folder (amavisd-new,spam...

Patch for new feature...
\ Keean Schupke (20 Nov 2007)
. \ Victor Duchovni (20 Nov 2007)
. . \ (Wietse Venema) (20 Nov 2007)
. . \ Keean Schupke (20 Nov 2007)
. . . \ (Wietse Venema) (20 Nov 2007)

4 msg	rsmtp and Postfix
7 msg	Earthlink rejecting based on account activity
7 msg	sender_relay configuration issues
13 msg	big problem
2 msg	Error with saslpasswd2
3 msg	what is wrong here?
5 msg	disable smtp_connection_cache_on_demand per des...
2 msg	Re: Digest of postfix-users list V1 #1865
3 msg	Bypass spamassassin check from trusted and alre...
3 msg	Postfix: safe workaround forsmtp_sasl_tls_verif...
6 msg	Configuring Postfix as SMTP Gateway for Outgoin...
3 msg	Postfix relaying depending on recipient address

Subject:	Re: Patch for new feature...
Group:	Postfix-users
From:	(Wietse Venema)
Date:	20 Nov 2007

Keean Schupke:
[ Charset ISO-8859-1 unsupported, converting... ]
> Hi,
> Thanks, the name was nothing special, so I have updated it, and I
> realised changing the resp->code was doing nothing as it is not kept,
> so I dropped it. I fixed the comment style too. I feel its probably
> better to only bounce an auth failure, as we might want to know about
> those other failures... My only concern would be if some mailservers
> returned unreliable error codes? I have attached an updated patch that
> still looks for a 535 error, but I am willing to go with whatever will
> get the change incoporated upstream, as it makes the maintainance of
> several mailservers much easier. The old version of the patch had been
> thoroughly tested - but these changes should not affect functionality.

I'll adopt this and then ignore the 35 portion of 535.

> I also have a question... can I get bounces to take a different route?

No. There are gazillion ways mail can fail. Making the bounce path
error-dependent would suck up development cycles until the end of
time.

Wietse

> For example if this new feature (smtp_sasl_auth_soft_bounce) is not
> enabled, the bounce messages go nowhere (as they are sent back to the
> same relay - where the authentication fails again). Effectively I want
> any soft failure from the relayhost to cause a mail to be queued and a
> hard failure to cause the bounce message to be sent directly, or via
> another relay. Is this currently possible?
>
> Regards
> Keean Schupke, Fry-IT Ltd.
>
> Here is the patch (v2):
>
> diff -r -U5 postfix-2.4.5.orig/src/global/mail_params.h
> postfix-2.4.5/src/global/mail_params.h
> --- postfix-2.4.5.orig/src/global/mail_params.h 2007-07-11
> 00:47:45.000000000 +0100
> +++ postfix-2.4.5/src/global/mail_params.h 2007-11-20
> 15:30:16.000000000 +0000
> @@ -1421,10 +1421,14 @@
> */
> #define VAR_SMTP_SASL_ENABLE "smtp_sasl_auth_enable"
> #define DEF_SMTP_SASL_ENABLE 0
> extern bool var_smtp_sasl_enable;
>
> +#define VAR_SMTP_SASL_AUTH_SOFT_BOUNCE "smtp_sasl_auth_soft_bounce"
> +#define DEF_SMTP_SASL_AUTH_SOFT_BOUNCE 0
> +extern bool var_smtp_sasl_auth_soft_bounce;
> +
> #define VAR_SMTP_SASL_PASSWD "smtp_sasl_password_maps"
> #define DEF_SMTP_SASL_PASSWD ""
> extern char *var_smtp_sasl_passwd;
>
> #define VAR_SMTP_SASL_OPTS "smtp_sasl_security_options"
> @@ -1516,10 +1520,14 @@
> */
> #define VAR_LMTP_SASL_ENABLE "lmtp_sasl_auth_enable"
> #define DEF_LMTP_SASL_ENABLE 0
> extern bool var_lmtp_sasl_enable;
>
> +#define VAR_LMTP_SASL_AUTH_SOFT_BOUNCE "lmtp_sasl_auth_soft_bounce"
> +#define DEF_LMTP_SASL_AUTH_SOFT_BOUNCE 0
> +extern bool var_lmtp_sasl_auth_soft_bounce;
> +
> #define VAR_LMTP_SASL_PASSWD "lmtp_sasl_password_maps"
> #define DEF_LMTP_SASL_PASSWD ""
> extern char *var_lmtp_sasl_passwd;
>
> #define VAR_LMTP_SASL_OPTS "lmtp_sasl_security_options"
> diff -r -U5 postfix-2.4.5.orig/src/smtp/smtp_params.c
> postfix-2.4.5/src/smtp/smtp_params.c
> --- postfix-2.4.5.orig/src/smtp/smtp_params.c 2006-10-05
> 17:37:40.000000000 +0100
> +++ postfix-2.4.5/src/smtp/smtp_params.c 2007-11-20
> 15:27:49.000000000 +0000
> @@ -86,10 +86,11 @@
> VAR_IGN_MX_LOOKUP_ERR, DEF_IGN_MX_LOOKUP_ERR, &var_ign_mx_lookup_err,
> VAR_SKIP_QUIT_RESP, DEF_SKIP_QUIT_RESP, &var_skip_quit_resp,
> VAR_SMTP_ALWAYS_EHLO, DEF_SMTP_ALWAYS_EHLO, &var_smtp_always_ehlo,
> VAR_SMTP_NEVER_EHLO, DEF_SMTP_NEVER_EHLO, &var_smtp_never_ehlo,
> VAR_SMTP_SASL_ENABLE, DEF_SMTP_SASL_ENABLE, &var_smtp_sasl_enable,
> + VAR_SMTP_SASL_AUTH_SOFT_BOUNCE,
> DEF_SMTP_SASL_AUTH_SOFT_BOUNCE, &var_smtp_sasl_auth_soft_bounce,
> VAR_SMTP_RAND_ADDR, DEF_SMTP_RAND_ADDR, &var_smtp_rand_addr,
> VAR_SMTP_QUOTE_821_ENV, DEF_SMTP_QUOTE_821_ENV, &var_smtp_quote_821_env,
> VAR_SMTP_DEFER_MXADDR, DEF_SMTP_DEFER_MXADDR, &var_smtp_defer_mxaddr,
> VAR_SMTP_SEND_XFORWARD, DEF_SMTP_SEND_XFORWARD, &var_smtp_send_xforward,
> VAR_SMTP_CACHE_DEMAND, DEF_SMTP_CACHE_DEMAND, &var_smtp_cache_demand,
> diff -r -U5 postfix-2.4.5.orig/src/smtp/smtp_sasl_glue.c
> postfix-2.4.5/src/smtp/smtp_sasl_glue.c
> --- postfix-2.4.5.orig/src/smtp/smtp_sasl_glue.c 2006-07-19
> 21:03:31.000000000 +0100
> +++ postfix-2.4.5/src/smtp/smtp_sasl_glue.c 2007-11-20
> 16:06:42.000000000 +0000
> @@ -338,10 +338,15 @@
>
> /*
> * We completed the authentication protocol.
> */
> if (resp->code / 100 != 2) {
> + /* begin: KPS20071120 - fix: SASL auth soft bounce */
> + if (var_smtp_sasl_auth_soft_bounce && resp->code == 535) {
> + STR(resp->dsn_buf)[0] = '4';
> + }
> + /* end: KPS20071120 - fix: SASL auth soft bounce */
> dsb_update(why, resp->dsn, DSB_DEF_ACTION,
> DSB_MTYPE_DNS, session->host,
> var_procname, resp->str,
> "SASL authentication failed; server %s said: %s",
> session->namaddr, resp->str);
>
>