[Bug 1833322] Re: Consider removing irqbalance from default install on desktop images

[Christian Ehrhardt ]

# Summary

This discussion was seeminly easier to make the more dedicated to a singluar
use case you are - as then you have less "but what if" cases to consider.
That wide usage is great for Ubuntu but sometimes delays decisions.

List of reasons to remove it from the default dependencies:
- Seems to cause issues more often on Desktop environments
- cpufreq, thermald and similar struggle to save energy
- Impacts due to unepexcted throttling
- Conflicts with enabling/disabling threads/cores
- Problematic in virtual environments
- It is mostly an x86 thing but we pull it in everywhere
- It conflicts with manually fine tuned IRQ affinity e.g. in
  ultra low latency setups
- It is less useful on cpus with large and wide shared caches
  as well as in virtual environments without fix pinning

List of reasons to keep it in the set of default dependencies:
- Benefits seem mostly for large scale servers
- lacking irqbalance can be a performance degradation in some
  large scale high traffic cases

I think from all I've found - old and new - it seems it still has its purpose
in some scenarios, but the HW/SW world evolved and it is nowadays less often
useful and more often harmful than it was in the past.
On the other hand there is almost no clear cut "it is bad and that is why",
most issues were individual issues and special cases, nothing that would
apply to everyone.

And irqbalance still has is purpose, so we should surely keep it around.

In a perfect worlds this would have half a year of time or more and two people
to run all kinds of workloads on all kinds of HW to compare. But let us be
honest that will not happen and that would then also be not be worth the effort.
We'll have to decide with what we have.
Have the others that switched have more time to evaluate in depth, I do not
know. But usually once a significant amount of the ecosystems changed and you
lack better data it is better to also follow or common hints and optimizations
will no more apply due to being the one outlier in regard to behavior.

To me this seems to be a perfect case for a few special images/deployments
known to match the workload profile that needs this to enable it.
It is also more likely that a professional admin of such a large scale machine
(or cluster thereof) can make the opt-in decision and evaluation better than
expectint every user of Ubuntu to think about an opt-out.


---

Options IMHO:
A) Change it from an opt-out to an opt-in and remove the dependency
   from ubuntu-standard
B) Remove it from ubuntu-standard to get rid of it in Desktops and images
   used in virtual environments. But try to keep it in a place that is mostly
   used for bare metal which tend to be closer to the kind that benefits more
C) Do nothing, keep it as is

D) Any of the above, but let us not touch Noble more than half way through the
cycle, but do that early in 24.10 to have enough exposure before a release in
an LTS.

My gut feeling (and it can't be much more without much more time for much
deeper investigations) would be (A).

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to ubuntu-meta in Ubuntu.
https://bugs.launchpad.net/bugs/1833322

Title:
  Consider removing irqbalance from default install on desktop images

Status in irqbalance package in Ubuntu:
  New
Status in ubuntu-meta package in Ubuntu:
  Confirmed

Bug description:
  as per https://github.com/pop-os/default-settings/issues/60

  Distribution (run cat /etc/os-release):

  $ cat /etc/os-release
  NAME="Pop!_OS"
  VERSION="19.04"
  ID=ubuntu
  ID_LIKE=debian
  PRETTY_NAME="Pop!_OS 19.04"
  VERSION_ID="19.04"
  HOME_URL="https://system76.com/pop"
  SUPPORT_URL="http://support.system76.com"
  BUG_REPORT_URL="https://github.com/pop-os/pop/issues"
  PRIVACY_POLICY_URL="https://system76.com/privacy"
  VERSION_CODENAME=disco
  UBUNTU_CODENAME=disco

  Related Application and/or Package Version (run apt policy $PACKAGE
  NAME):

  $ apt policy irqbalance
  irqbalance:
  Installed: 1.5.0-3ubuntu1
  Candidate: 1.5.0-3ubuntu1
  Version table:
  *** 1.5.0-3ubuntu1 500
  500 http://us.archive.ubuntu.com/ubuntu disco/main amd64 Packages
  100 /var/lib/dpkg/status

  $ apt rdepends irqbalance
  irqbalance
  Reverse Depends:
  Recommends: ubuntu-standard
  gce-compute-image-packages

  Issue/Bug Description:

  as per konkor/cpufreq#48 and
  http://konkor.github.io/cpufreq/faq/#irqbalance-detected

  irqbalance is technically not needed on desktop systems (supposedly it
  is mainly for servers), and may actually reduce performance and power
  savings. It appears to provide benefits only to server environments
  that have relatively-constant loading. If it is truly a server-
  oriented package, then it shouldn't be installed by default on a
  desktop/laptop system and shouldn't be included in desktop OS images.

  Steps to reproduce (if you know):

  This is potentially an issue with all default installs.

  Expected behavior:

  n/a

  Other Notes:

  I can safely remove it via "sudo apt purge irqbalance" without any
  apparent adverse side-effects. If someone is running a situation where
  they need it, then they always have the option of installing it from
  the repositories.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/irqbalance/+bug/1833322/+subscriptions