On 10/09/2010 05:14 PM, Reindl Harald wrote: > I think this is a bug on dbmail-side > > lmtpd should send a 4xx-temorary-error instead a 5xx-hard-bounce > So postfix would try later to deliver the message again
That is easier said than done. The current setup will put lmtpd into sleep-mode immediately if the database goes away, and stay there until the connection recovers. In sleep mode no new client connections are accepted, and indeed in a perfect world existing connections should be dealt with gracefully. So obviously there are race-conditions at play here. Currently queries either succeed, or queries fail. The exact nature of the query failure is *not* examined, nor communicated back up the call stack! All query errors are hard errors. The solution in this case is simple of course: do not shut down your database, unless you've shut down postfix or dbmail-lmtpd first. In fact, shutting down lmtpd during dbmail-maintenance runs doesn't sound too bad. > > Postfix does this as sample if he losts the connection to lmtp for whatever > reason, but not if the target gives a 5xx-error back > > Am 09.10.2010 17:06, schrieb Dis McCarthy: >> I got a bounce message (to the same alias that failed, strangely enough) >> during the nightly cron run. It looks >> like several messages were outright discarded at that time. Shouldn't a db >> connection reset (or db unavailable) be >> a temporary error? Is there a place to configure that? >> >> <[email protected]> (expanded from <root>): host 127.0.0.1[127.0.0.1] said: >> 550 >> Recipient <[email protected]> FAIL (in reply to RCPT TO command) >> >> Looking at the logs: >> >> Oct 9 05:00:03 mayhem postfix/pickup[10082]: 309771C0405: uid=0 from=<root> >> Oct 9 05:00:03 mayhem postfix/cleanup[12188]: 309771C0405: >> message-id=<[email protected]> >> Oct 9 05:00:03 mayhem postfix/qmgr[21315]: 309771C0405: >> from=<[email protected]>, size=763, nrcpt=1 (queue active) >> Oct 9 05:00:03 mayhem dbmail/lmtpd[20778]: Message:[serverchild] >> serverchild.c,PerformChildTask(+349): incoming connection from [127.0.0.1] >> by pid [20778] >> Oct 9 05:00:03 mayhem dbmail/lmtpd[20778]: Error:[sql] >> dbpgsql.c,db_query(+287): query failed [SELECT deliver_to FROM >> dbmail_aliases WHERE lower(alias) = lower('[email protected]') AND >> lower(alias) <> lower(deliver_to)] : [FATAL: terminating connection due to >> administrator command#012server closed the connection >> unexpectedly#012#011This probably means the server terminated >> abnormally#012#011before or while processing the request.#012] >> Oct 9 05:00:03 mayhem dbmail/lmtpd[20778]: Error:[auth] >> authsql.c,__auth_query(+293): error executing query >> Oct 9 05:00:03 mayhem postfix/lmtp[12191]: 309771C0405: >> to=<[email protected]>, orig_to=<root>, relay=127.0.0.1[127.0.0.1]:24, >> delay=0.73, delays=0.46/0.02/0.07/0.18, dsn=5.0.0, status=bounced (host >> 127.0.0.1[127.0.0.1] said: 550 Recipient <[email protected]> FAIL (in reply >> to RCPT TO command)) >> Oct 9 05:00:03 mayhem postfix/cleanup[12188]: 94A9C1C041B: >> message-id=<[email protected]> >> Oct 9 05:00:03 mayhem postfix/bounce[12193]: 309771C0405: sender >> non-delivery notification: 94A9C1C041B >> >> >> >> >> _______________________________________________ >> DBmail mailing list >> [email protected] >> http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail > > > > _______________________________________________ > DBmail mailing list > [email protected] > http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail -- ________________________________________________________________ Paul Stevens paul at nfg.nl NET FACILITIES GROUP GPG/PGP: 1024D/11F8CD31 The Netherlands________________________________http://www.nfg.nl _______________________________________________ DBmail mailing list [email protected] http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail
