summaryrefslogtreecommitdiffstats
path: root/RELEASE_NOTES-2.9
diff options
context:
space:
mode:
Diffstat (limited to 'RELEASE_NOTES-2.9')
-rw-r--r--RELEASE_NOTES-2.9352
1 files changed, 352 insertions, 0 deletions
diff --git a/RELEASE_NOTES-2.9 b/RELEASE_NOTES-2.9
new file mode 100644
index 0000000..e30a34d
--- /dev/null
+++ b/RELEASE_NOTES-2.9
@@ -0,0 +1,352 @@
+The stable Postfix release is called postfix-2.9.x where 2=major
+release number, 9=minor release number, x=patchlevel. The stable
+release never changes except for patches that address bugs or
+emergencies. Patches change the patchlevel and the release date.
+
+New features are developed in snapshot releases. These are called
+postfix-2.10-yyyymmdd where yyyymmdd is the release date (yyyy=year,
+mm=month, dd=day). Patches are never issued for snapshot releases;
+instead, a new snapshot is released.
+
+The mail_release_date configuration parameter (format: yyyymmdd)
+specifies the release date of a stable release or snapshot release.
+
+If you upgrade from Postfix 2.7 or earlier, read RELEASE_NOTES-2.8
+before proceeding.
+
+Major changes - critical
+------------------------
+
+[Incompat 20110321] You need to "postfix reload" after upgrade from
+snapshot 20110320 or earlier. The hash_queue_names algorithm was
+changed to provide better performance with long queue IDs.
+
+[Incompat 20110313] Use "postfix reload" after "make upgrade" on a
+running Postfix system. This is needed because the protocol between
+postscreen(8) and dnsblog(8) has changed.
+
+Major changes - library API
+---------------------------
+
+[Incompat 20110130] The VSTREAM error flags are now split into
+separate read and write error flags. As a result of this change,
+all programs that use Postfix VSTREAMs MUST be recompiled.
+
+Major changes - compatibility
+-----------------------------
+
+[Incompat 20111012] For consistency with the SMTP standard, the
+(client-side) smtp_line_length_limit default value was increased
+from 990 characters to 999 (i.e. 1000 characters including <CR><LF>).
+Specify "smtp_line_length_limit = 990" to restore historical Postfix
+behavior.
+
+[Incompat 20111012] To simplify integration with third-party
+applications, the Postfix sendmail command now always transforms
+all input lines ending in <CR><LF> into UNIX format (lines ending
+in <LF>). Specify "sendmail_fix_line_endings = strict" to restore
+historical Postfix behavior (i.e. convert all input lines ending
+in <CR><LF> only if the first line ends in <CR><LF>).
+
+[Incompat 20111106] To work around broken remote SMTP servers, the
+Postfix SMTP client by default no longer appends the "AUTH=<>"
+option to the MAIL FROM command. Specify "smtp_send_dummy_mail_auth
+= yes" to restore the old behavior.
+
+Major changes - gradual degradation
+-----------------------------------
+
+[Incompat 20120114] Logfile-based alerting systems may need to be
+updated to look for "error" messages in addition to "fatal" messages.
+Specify "daemon_table_open_error_is_fatal = yes" to get the historical
+behavior (immediate termination with "fatal" message).
+
+[Feature 20120114] Instead of terminating immediately with a "fatal"
+message when a database file can't be opened, a Postfix daemon
+program now logs an "error" message, and continues execution with
+reduced functionality. For the sake of sanity, the number of
+"errors" over the life of a process is limited to 13.
+
+Features that don't depend on the unavailable table will continue
+to work; attempts to use features that depend on the table will
+fail, and will be logged with a "warning" message.
+
+[Feature 20120108] Instead of terminating with a fatal error, the
+LDAP, *SQL and memcache clients now handle table lookup errors in
+the "domain" feature, instead of terminating with a fatal error.
+
+[Feature 20120102] Degrade gradually when some or all network
+protocols specified with inet_protocols are unavailable, instead
+of terminating with a fatal error. This eliminates build errors on
+non-standard systems where opening an IPv4 socket results in an
+error, and on non-standard systems where opening an IPv6 socket
+results in an error. In the worst case, the master daemon will log
+a message that it disables all type "inet" services. This will still
+allow local submission and local delivery.
+
+[Feature 20111222] Instead of terminating with a fatal error, the
+Postfix SMTP server now handles errors with database lookups in
+mynetworks, TLS client certificate tables, debug_peer_list,
+smtpd_client_event_limit_exceptions, permit_mx_backup_networks and
+local_header_rewrite_clients, and reports "server local data error"
+or "temporary lookup error".
+
+[Feature 20111229] Instead of terminating with a fatal error, the
+trivial-rewrite server now handles errors with database lookups in
+virtual_alias_domains, relay_domains, virtual_mailbox_domains. This
+means fewer occasions where trivial-rewrite clients (such as the
+SMTP server) will appear to hang.
+
+Major changes - long queue IDs
+------------------------------
+
+Postfix 2.9 introduces support for non-repeating queue IDs (also
+used as queue file names). These names are encoded in a mix of upper
+case, lower case and decimal digit characters. Long queue IDs are
+disabled by default to avoid breaking tools that parse logfiles and
+that expect queue IDs with the smaller [A-F0-9] character set.
+
+[Incompat 20110320] If you enable support for long queue file names,
+you need to be aware that these file names are not compatible with
+Postfix <= 2.8. If you must migrate back to Postfix <= 2.8, you
+must first convert all long queue file names into short names,
+otherwise the old Postfix version will complain.
+
+The conversion procedure before migration to Postfix <= 2.8 is:
+
+ # postfix stop
+ # postconf enable_long_queue_ids=no
+ # postsuper
+
+Run the postsuper command repeatedly until it no longer reports
+queue file name changes.
+
+[Feature 20110320] Support for long, non-repeating, queue IDs (queue
+file names). The benefit of non-repeating names is simpler logfile
+analysis, and easier queue migration (if you don't merge different
+queues, there is no need to run "postsuper" to change queue file
+names that don't match their message file inode number).
+
+Specify "enable_long_queue_ids = yes" to enable the feature. This
+does not change the names of existing queue files. See postconf(5)
+or postconf.5.html#enable_long_queue_ids for a detailed description
+of the differences with the old short queue IDs.
+
+This changes new Postfix queue IDs from the short form 0FCEE9247A9
+into the longer form 3Ps0FS1Zhtz1PFjb, and changes new Message-ID
+header values from YYMMDDHHMMSS.queueid@myhostname into the shorter
+form queueid@myhostname.
+
+Major changes - memcache
+------------------------
+
+[Feature 20111209] memcache lookup and update support. This provides
+a way to share postscreen(8) or verify(8) caches between Postfix
+instances. See MEMCACHE_README and memcache_table(5) for details
+and limitations.
+
+[Feature 20111213] Support for a persistent backup database in the
+memcache client. The memcache client updates the memcache whenever
+it looks up or modifies information in the persistent database.
+
+Major changes - postconf
+------------------------
+
+The postconf command was restructured - it now warns about unused
+parameter name=value settings in main.cf or master.cf (likely to
+be mistakes), it now understands "dynamic" parameter names such as
+parameters whose name depends on the name of a master.cf entry, and
+it can display main.cf and master.cf in a more user-friendly format.
+
+[Feature 20120117] support for legacy database parameter names
+(main.cf parameter names that are generated by prepending a suffix
+to the database name).
+
+[Feature 20111118] The "postconf -M" (display master.cf) command
+now supports filtering. For example, specify "postconf -M inet"
+to display only services that listen on the network.
+
+[Feature 20111113] postconf support to warn about unused "name=value"
+entries in main.cf, and about unused "-o name=value" entries in
+master.cf. This should help to eliminate common errors with mis-typed
+names.
+
+[Feature 20111108] postconf support for parameter names that are
+generated automatically from master.cf entries (delivery agents,
+spawn services), and for parameter names that are defined with
+main.cf smtpd_restriction_classes.
+
+[Feature 20111106] "postconf -M" support to print master.cf entries,
+and "postconf -f" support to fold long main.cf or master.cf lines
+for human readability.
+
+Major changes - trickle defense
+-------------------------------
+
+[Feature 20110212] Support for per-record deadlines. These change
+the behavior of Postfix timeout parameters, from a time limit per
+read or write system call, to a time limit to send or receive a
+complete record (an SMTP command line, SMTP response line, SMTP
+message content line, or TLS protocol message). This limits the
+impact from hostile peers that trickle data one byte at a time.
+
+The new configuration parameters and their default settings are:
+smtpd_per_record_deadline (normal: no, overload: yes),
+smtp_per_record_deadline (no), and lmtp_per_record_deadline (no).
+
+Note: when per-record deadlines are enabled, a short time limit may
+cause problems with TLS over very slow network connections. The
+reason is that a TLS protocol message can be up to 16 kbytes long
+(with TLSv1), and that an entire TLS protocol message must be sent
+or received within the per-record deadline.
+
+Per-record deadlines were introduced with postscreen(8) in Postfix
+2.8. This program does not receive mail, and therefore it has no
+problems with TLS over slow connections.
+
+Major changes - postscreen
+--------------------------
+
+[Feature 20111211] The proxymap(8) server can now be used to share
+postscreen(8) or verify(8) caches between Postfix instances. Support
+for proxymap-over-TCP, to share a Postfix database between hosts,
+is expected to be completed in the Postfix 2.10 development cycle.
+
+[Feature 20111209] memcache lookup and update support. This provides
+a way to share postscreen(8) or verify(8) caches between Postfix
+instances.
+
+[Feature 20110228] postscreen(8) support to force remote SMTP clients
+to implement proper MX lookup policy. By listening on both primary
+and backup MX addresses, postscreen(8) can deny the temporary
+whitelist status to clients that connect only to backup MX hosts,
+and prevent them from talking to a Postfix SMTP server process.
+
+Example: when 1.2.3.4 is a local backup IP address, specify
+"postscreen_whitelist_interfaces = !1.2.3.4 static:all".
+
+Major changes - tls
+-------------------
+
+[Incompat 20111205] Postfix now logs the result of successful TLS
+negotiation with TLS logging levels of 0. See the smtp_tls_loglevel
+and smtpd_tls_loglevel descriptions in the postconf(5) manpage for
+other minor differences.
+
+[Feature 20111205] Support for TLS public key fingerprint matching
+in the Postfix SMTP client (in smtp_tls_policy_maps) and server (in
+check_ccert access maps). Public key fingerprints are inherently
+more specific than fingerprints over the entire certificate.
+
+[Feature 20111205] Revision of Postfix TLS logging. The main
+difference is that Postfix now logs the result of successful TLS
+negotiation with TLS logging levels of 0. See the smtp_tls_loglevel
+and smtpd_tls_loglevel descriptions in the postconf(5) manpage for
+other minor differences.
+
+Major changes - sasl authentication
+-----------------------------------
+
+[Incompat 20111218] To support external SASL authentication, e.g.,
+in an NGINX proxy daemon, the Postfix SMTP server now always checks
+the smtpd_sender_login_maps table, even without having
+"smtpd_sasl_auth_enable = yes" in main.cf.
+
+[Feature 20111218] Support for external SASL authentication via the
+XCLIENT command. This is used to accept SASL authentication from
+an SMTP proxy such as NGINX. This support works even without having
+to specify "smtpd_sasl_auth_enable = yes" in main.cf.
+
+[Incompat 20111106] To work around broken remote SMTP servers, the
+Postfix SMTP client by default no longer appends the "AUTH=<>"
+option to the MAIL FROM command. Specify "smtp_send_dummy_mail_auth
+= yes" to restore the old behavior.
+
+Major changes - large file support
+----------------------------------
+
+[Feature 20110219] Postfix now uses long integers for message_size_limit,
+mailbox_size_limit and virtual_mailbox_limit. On LP64 systems (64-bit
+long and pointer, but 32-bit integer), these limits can now exceed
+2GB.
+
+Major changes - ipv6
+--------------------
+
+[Incompat 20110918] The following changes were made in default
+settings, in preparation for general availability of IPv6:
+
+- The default inet_protocols value is now "all" instead of "ipv4",
+ meaning use both IPv4 and IPv6.
+
+ To avoid an unexpected loss of performance for sites without
+ global IPv6 connectivity, the commands "make upgrade" and "postfix
+ upgrade-configuration" now append "inet_protocols = ipv4" to
+ main.cf when no explicit inet_protocols setting is already present.
+ This workaround will be removed in a future release.
+
+- The default smtp_address_preference value is now "any" instead
+ of "ipv6", meaning choose randomly between IPv6 and IPv4. With
+ this the Postfix SMTP client will have more success delivering
+ mail to sites that have problematic IPv6 configurations.
+
+Major changes - address verification
+------------------------------------
+
+[Feature 20111211] The proxymap(8) server can now be used to share
+postscreen(8) or verify(8) caches between Postfix instances. Support
+for proxymap-over-TCP, to share a Postfix database between hosts,
+is expected to be completed in the Postfix 2.10 development cycle.
+
+[Feature 20111209] memcache lookup and update support. This provides
+a way to share postscreen(8) or verify(8) caches between Postfix
+instances.
+
+[Feature 20111203] Support for time-dependent sender addresses
+of address verification probes. The default address, double-bounce,
+may end up on spammer blacklists. Although Postfix discards mail
+for this address, such mail still uses up network bandwidth and
+server resources. Specify an address_verify_sender_ttl value of
+several hours or more to frustrate address harvesting.
+
+Major changes - session transcript notification
+-----------------------------------------------
+
+[Incompat 20120114] By default the Postfix SMTP server no longer
+reports transcripts of sessions where a client command is rejected
+because a lookup table is unavailable. Postfix now implements gradual
+degradation, for example, the SMTP server keeps running instead of
+terminating with a fatal error. This change in error handling would
+result in a very large number of "transcript of session" email
+notifications when an LDAP or *SQL server goes down).
+
+To receive such reports, add the new "data" class to the notify_classes
+parameter value. The reports will be sent to the error_notice_recipient
+address as before. This class is also used by the Postfix SMTP
+client to report about sessions that fail because a table is
+unavailable.
+
+Major changes - logging
+----------------------------------------
+
+[Incompat 20120114] Logfile-based alerting systems may need to be
+updated to look for "error" messages in addition to "fatal" messages.
+Specify "daemon_table_open_error_is_fatal = yes" to get the historical
+behavior (immediate termination with "fatal" message).
+
+[Incompat 20111214] Logfile-based analysis tools may need to be
+updated. The submission and smtps examples in the sample master.cf
+file were updated to make their logging easier to distinguish.
+
+See the source file pflogsumm_quickfix.txt for a "quick fix".
+
+[Incompat 20111205] Postfix now logs the result of successful TLS
+negotiation with TLS logging levels of 0. See the smtp_tls_loglevel
+and smtpd_tls_loglevel descriptions in the postconf(5) manpage for
+other minor differences.
+
+[Incompat 20110219] The Postfix SMTP and QMQP servers now log
+"hostname X does not resolve to address Y", when a "reverse hostname"
+lookup result does not resolve to the client IP address. Until now
+these servers logged "Y: hostname X verification failed" or "Y:
+address not listed for hostname X" which people found confusing.