On Wed, Nov 13, 2013 at 08:34:51AM -0800, boson code wrote: > Whenever I try to send an email from my server, I get the following error: > > Nov 13 06:37:21 xyz postfix/smtpd[6730]:connect from unknown[a.b.c.d] > Nov 13 06:37:21 xyz postfix/smtp[6729]: warning: host X.com[x.y.z.d]:25 > greeted me with my own hostname xyz.biz > Nov 13 06:37:21 xyz postfix/smtp[6729]: warning: host X.com[x.y.z.d]:25 > replied to HELO/EHLO with my own hostname xyz.biz > Nov 13 06:37:21 xyz postfix/smtp[6729]: 2017F1B00C54: to=<du...@x.com>, > relay=X.com[x.y.z.d]:25, delay=0.98, delays=0.17/0/0.81/0, dsn=5.4.6, > status=bounced (mail for X.com loops back to myself)
Obfuscating the IP addresses and domains makes it very difficult to help you. The recipient domain was not configured for local delivery, and yet its best MX host appears to be your system. See http://www.postfix.org/BASIC_CONFIGURATION_README.html Below your configuration is almost entirely a misguided mess of conflicting an unwise settings. Throw it all away and start from scratch with something small that changes just a few parameters (mydestination, mynetworks, ...) and leaves the rest at default values. Make that work, and DO NOT randomly tweak settings you don't understand. > address_verify_poll_delay = 1s This is too aggressive. > body_checks_size_limit = 40980000 > receive_override_options = no_header_body_checks While this is a waste of CPU, assuming body checks actually take place via master.cf overrides... > connection_cache_ttl_limit = 300000s > smtp_connection_cache_time_limit = 30000s > smtp_connection_reuse_time_limit = 30000s This is inconsiderate ABUSE of remote system resources. Restore these settings back to default values. > default_delivery_slot_cost = 2 > default_minimum_delivery_slots = 3 > default_recipient_refill_delay = 1s > default_recipient_refill_limit = 10 > smtp_delivery_slot_cost = 2 > smtp_minimum_delivery_slots = 1 > smtp_recipient_refill_delay = 1s > smtp_recipient_refill_limit = 1000 You probably should not be changing these. > default_destination_concurrency_limit = 10 > default_destination_recipient_limit = 1 This is a terrible idea. Concurrency is now counted per-recipient, not per-destinatin. > default_process_limit = 10 > qmgr_message_active_limit = 500 > qmgr_message_recipient_limit = 500 > qmgr_message_recipient_minimum = 1 Are you running on a Raspberry Pi? > hash_queue_depth = 2 And yet building up very large queues? > service_throttle_time = 1s This is crazy, reset back to approximately default value. > smtp_connect_timeout = 1s This is way too aggressive for general use. Are you isolated from the Internet? > smtp_destination_concurrency_limit = 10000 > smtp_destination_recipient_limit = 1 How on earth is that Raspberry Pi managing 10000 concurrent connections to each individual recipient? > smtpd_delay_open_until_valid_rcpt = no > smtpd_delay_reject = no Unwise. -- Viktor.
