Hi,

HAProxy 2.3.10 was released on 2021/04/23. It added 33 new commits after
version 2.3.9.

The most notable fixes of this release are about the H2 multiplexer. Two
bugs were fixed in the CONTINUATION frames parsing. The first one was about
wakeups in loop of the h2 connections if only part of a CONTINUATION frame
header was received into a full demux buffer. In this case, we were unable
to detect the frames were too large to be processed. The H2 connection was
infinitely waiting for more data with a full demux buffer and was woken up
in loops to process already received data. While it is a pretty annoying bug
because it consumes all idle CPU for nothing, it is not a DOS bug because
incoming traffic is still processed. The second bug was about the frames
size calculation when CONTINUATION and HEADERS frames were merged. Only
available data were counted instead of the total frames size. The last bug
on the H2 multiplexer was about the shutdowns detection. When a shutdown was
received along with data, it was ignored. Thus, in this case, the H2
connection remained blocked until its timeout expiration.

Here is the list of other fixes for this release:

  - Amaury fixed a bug in the cpu-map notation when both processes and
    threads were specified, most specifically P-Q/1 or 1/P-Q notation.

  - The possible deadlock under thread isolation of processes built with
    DEBUG_UAF, supposed to be fixed in a previous release was not
    totally. It is fixed now.

  * Thayne McCombs fixed the field converter. The size of the sample buffer
    was not properly updated. He also fixed the parsing of "us" unit for
    timers.

  * fc_http_major and bc_http_major sample fetches were fixed to fail when
    called for connections attached to a non-HTX multiplexer. In addition it
    is now possible to call them from the health-checks.

  * The support of log-format strings in a health-check context was
    improved. It is now possible to use backend and server related formats
    (%s, %sc, %b, %bi, %bp, %si and %sp). To make it possible, an identifier
    was set on the dummy frontend used for health-checks.

  * Several SSL sample fetches (ssl_bc_*) were fixed to work when called in
    a health-check context.

  * The silent-drop action was fixed to work for IPv6 when permissions are
    insufficient.

  * H1 idle connections on server side are now closed if some data are
    received. This avoids to send pending 408-Request-time-out responses to
    clients. Note it is still possible to receive a 408 response from a
    server if it is received in same time the request is sent.

  * The true number of retries is now reported in logs if no connection was
    attempted. Since the beginning, when the session was aborted before any
    connection attempt to any server, the backend retries value was
    reported, instead of 0.

  * BUG_ON() statements were removed from http_get_stline() function. This
    function already returns NULL when it fails to get the start-line from
    an HTX message. There is no reason to restrict it, especially because
    there are some valid cases to call it on an empty buffer or after data
    forwarding.

  * IP parsing in hdr_ip() sample fetch is now stricter. It was fixed to
    never return invalid IP addresses.

  * William fixed two bugs in the master-worker affecting the peers
    synchronization. Still about peers, Emeric fixed bugs during reloads
    preventing the peers synchronization to work properly.

  * The update on the CLI of the default SSL certificate used not to work
    correctly as the previous one was not removed, resulting in a random
    behavior namely on the SNI.

  * The method sample fetch was fixed to be sure to start-line is still
    there when an exotic method is found. For default methods (GET, POST..),
    transaction flags are used. But for unknown methods, we get it from the
    HTX message. In this case, the start-line must be there.

Thanks everyone for your help and your contributions !

Please find the usual URLs below :
   Site index       : http://www.haproxy.org/
   Discourse        : http://discourse.haproxy.org/
   Slack channel    : https://slack.haproxy.org/
   Issue tracker    : https://github.com/haproxy/haproxy/issues
   Wiki             : https://github.com/haproxy/wiki/wiki
   Sources          : http://www.haproxy.org/download/2.3/src/
   Git repository   : http://git.haproxy.org/git/haproxy-2.3.git/
   Git Web browsing : http://git.haproxy.org/?p=haproxy-2.3.git
   Changelog        : http://www.haproxy.org/download/2.3/src/CHANGELOG
   Cyril's HTML doc : http://cbonte.github.io/haproxy-dconv/


---
Complete changelog :
Amaury Denoyelle (2):
      BUG/MINOR: server: free srv.lb_nodes in free_server
      BUG/MEDIUM: config: fix cpu-map notation with both process and threads

Christopher Faulet (15):
      DOC: Explicitly state only IPv4 are supported by forwardfor/originalto 
options
      MINOR: No longer rely on deprecated sample fetches for predefined ACLs
      BUG/MEDIUM: threads: Ignore current thread to end its harmless period
      BUG/MINOR: checks: Set missing id to the dummy checks frontend
      MINOR: logs: Add support of checks as session origin to format lf strings
      BUG/MINOR: connection: Fix fc_http_major and bc_http_major for TCP 
connections
      MINOR: connection: Make bc_http_major compatible with tcp-checks
      BUG/MINOR: ssl-samples: Fix ssl_bc_* samples when called from a 
health-check
      BUG/MINOR: http-fetch: Make method smp safe if headers were already 
forwarded
      BUG/MINOR: http_htx: Remove BUG_ON() from http_get_stline() function
      BUG/MINOR: logs: Report the true number of retries if there was no 
connection
      BUG/MINOR: mux-h1: Release idle server H1 connection if data are received
      BUG/MAJOR: mux-h2: Properly detect too large frames when decoding headers
      BUG/MEDIUM: mux-h2: Fix dfl calculation when merging CONTINUATION frames
      BUG/MEDIUM: mux-h2: Properly handle shutdowns when received with data

Emeric Brun (3):
      BUG/MINOR: peers: remove useless table check if initial resync is finished
      BUG/MEDIUM: peers: re-work connection to new process during reload.
      BUG/MEDIUM: peers: re-work refcnt on table to protect against flush

Julien Pivotto (1):
      DOC: clarify that compression works for HTTP/2

Remi Tricot-Le Breton (4):
      BUG/MINOR: ssl: Fix update of default certificate
      BUG/MINOR: ssl: Prevent removal of crt-list line if the instance is a 
default one
      BUG/MINOR: ssl: Add missing free on SSL_CTX in ckch_inst_free
      DOC: ssl: Certificate hot update only works on fronted certificates

Thayne McCombs (2):
      BUG/MINOR: tools: fix parsing "us" unit for timers
      BUG/MEDIUM: sample: Fix adjusting size in field converter

William Lallemand (4):
      REGTESTS: ssl: "set ssl cert" and multi-certificates bundle
      REGTESTS: ssl: mark set_ssl_cert_bundle.vtc as broken
      BUG/MINOR: mworker/init: don't reset nb_oldpids in non-mworker cases
      BUG/MINOR: mworker: don't use oldpids[] anymore for reload

Willy Tarreau (5):
      BUILD: backend: fix build breakage in idle conn locking fix
      BUG/MINOR: tcp: fix silent-drop workaround for IPv6
      BUILD: tcp: use IPPROTO_IPV6 instead of SOL_IPV6 on FreeBSD/MacOS
      BUG/MINOR: http_fetch: make hdr_ip() resistant to empty fields
      CONTRIB: halog: fix issue with array of type char

--
Christopher Faulet

Reply via email to