← Back to team overview

debcrafters-packages team mailing list archive

[Bug 1799265] Autopkgtest regression report (avahi/0.8-5ubuntu5.3)

 

All autopkgtests for the newly accepted avahi (0.8-5ubuntu5.3) for jammy have finished running.
The following regressions have been reported in tests triggered by the package:

libreoffice/1:7.3.7-0ubuntu0.22.04.10 (arm64, s390x)
nss-mdns/0.15.1-1ubuntu1 (s390x)
pappl/1.0.3-2 (arm64)


Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-
migration/jammy/update_excuses.html#avahi

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

-- 
You received this bug notification because you are a member of
Debcrafters packages, which is subscribed to avahi in Ubuntu.
https://bugs.launchpad.net/bugs/1799265

Title:
  avahi-daemon high cpu, unusable networking

Status in Avahi:
  New
Status in avahi package in Ubuntu:
  Fix Released
Status in avahi source package in Jammy:
  Fix Committed
Status in avahi source package in Noble:
  Fix Released

Bug description:
  Currently running Kubuntu 18.10, Dell XPS 13 9350

  Since updating from Kubuntu 18.04 to 18.10, the avahi-daemon has been
  consistently hampering network performance and using CPU for long
  periods of time.

  When booting machine from off state, avahi-daemon uses an entire CPU
  at max load for approx 10 minutes. During this time, internet
  connectivity via wifi is essentially unusable. The wifi connection is
  good, but it seems that http transactions are cutoff mid-way so no
  webpage is able to load.

  When waking from sleep, the avahi-daemon causes similar symptoms, but
  with less than 1 full cpu usage, and with somewhat less degraded
  network performance, but still quite unusable.

  I have never had issues with avahi prior to the 18.10 upgrade, so I am
  fairly confident the issue is rooted in that change.

  ProblemType: Bug
  DistroRelease: Ubuntu 18.10
  Package: avahi-daemon 0.7-4ubuntu2
  ProcVersionSignature: Ubuntu 4.18.0-10.11-generic 4.18.12
  Uname: Linux 4.18.0-10-generic x86_64
  ApportVersion: 2.20.10-0ubuntu13
  Architecture: amd64
  CurrentDesktop: KDE
  Date: Mon Oct 22 10:00:34 2018
  InstallationDate: Installed on 2017-07-24 (455 days ago)
  InstallationMedia: Ubuntu 17.04 "Zesty Zapus" - Release amd64 (20170412)
  ProcEnviron:
   PATH=(custom, no user)
   XDG_RUNTIME_DIR=<set>
   LANG=en_US.UTF-8
   LD_PRELOAD=<set>
   SHELL=/bin/bash
  SourcePackage: avahi
  UpgradeStatus: Upgraded to cosmic on 2018-10-20 (2 days ago)

  [ Impact ]

   * avahi-daemon consumes a large amount of CPU over time, particularly
  on laptop/desktop use cases. Many reports of it burning 100% CPU.

   * This also leads to increased multicast traffic which has been
  observed to cause high packet loss on WiFi networks at very large
  events with many Ubuntu laptops on the same network.

   * This issue was fixed long ago fixed upstream, in Debian and in
  Kinetic+ but not previously backported to Jammy.

  [ Test Plan ]

  The problem with this bug is that when a watch is cleaned up, it stops
  timeouts being cleaned up. The list of timeouts then grows and when
  iterated causes high CPU usage.

  Watches are created when nss-mdns connects via the simple socket, so
  we can trigger this with an DNS lookup for a .local host: getent hosts
  avahitest1.local

  Timeouts are created when a DBus client disconnects, so we can trigger
  this by calling avahi-browse -a -t which will print out all services
  and then terminate the client.

  1. Launch two test VMs, avahitest1 and avahitest2. Install avahi-
  utils, avahi-daemon, libnss-mdns and sysstat. Upgrade to the fix on
  one of them only.

  2. Launch mpstat 15 on both hosts

  3. Launch many parallel avahi-browse on both: yes | xargs -n1 -I{}
  -P20 avahi-browse -a -t

  4. Launch many DNS lookups on both:
     while true; do getent hosts avahitest2.local; sleep 0.1; done

  5. Observe over time the CPU usage of one VM slowly grows, while the
  other does not.

  [ Where problems could occur ]

   * This patch is a single line change for a missing cleanup, that has
  shipped in Kinetic+ for several years

   * The version shipping in those releases is substantially similar to
  the current release (no new upstream release)

   * However: avahi-daemon is a default package on Desktops, so there is
  a high potential impact

  [ Other Info ]

   * Originally diagnosed/resolved in the Debian bug: 
   https://bugs-devel.debian.org/cgi-bin/bugreport.cgi?bug=993051

   * Upstream PR: https://github.com/avahi/avahi/pull/366

   * This patch ships in later versions (even the current version),
  however I renamed the patch to prefix the LP bug number per the
  preferences, and added DEP3 headers.

To manage notifications about this bug go to:
https://bugs.launchpad.net/avahi/+bug/1799265/+subscriptions