Belated response, but just for the record, Paride's recounting of upstream's
position in the context of the Debian decision was definitive for me:
On Wed, Jan 10, 2024 at 11:47:56AM -, Paride Legovini wrote:
> Back in the day I asked upstream their take on irqbalance usefulness
> with newer
** Changed in: ubuntu-z-systems
Status: New => Confirmed
--
You received this bug notification because you are a member of Desktop
Packages, which is subscribed to ubuntu-meta in Ubuntu.
https://bugs.launchpad.net/bugs/1833322
Title:
Consider removing irqbalance from default install on
Speaking from a Desktop perspective, it's difficult to have a strong
opinion without data to backup the decision but it does feel like that
in the light of what other distributions/upstream are doing we should
reverse the default and go with option A and not have it by default but
an opt-in
Hi Christian,
Thank you, yes I don't disagree with anything you said. There can be no
"one size fits all" and customizing performance tuning will always be
important but I will argue
1. There can be a "one size fits most" at least for desktop client
environments ("general optimization")
2. It may
Hi Etanay,
I realize I maybe wrote too much :-/
So I start with a TL;DR:
AFAICS you are right in all you say, but I think there can not be "one right
answer" anyway. Hence I'm trying to leave all parties their freedom of defining
what is important to them and try to learn from them what impact
Hi Christian,
Thank you. Yes I was not arguing strictly against irqbalance, just trying
to ascertain some discussion parameters as well as parameters for data
collection.
I have not yet seen a coherent philosophy on what it means to "optimize
performance" with default settings that serve the
Hi Christian,
Thank you for your reply to my post.
> I do not know the ping pong test,
A simple token passing ring, that is useful for getting the system to utilize
shallow idles states. Otherwise it can be difficult to get to such shallow
states on my test system, without them being timer
Just saw a mention of this bug, and I wanted to provide another
datapoint: I recently sponsored a SRU for an irqbalance bugfix (LP
#2038300), it was for an edge server platform (NVIDIA IGX Orin). What I
noticed was that the code was inherently racy and hard to validate with
unit tests because it's
** Tags added: architecture-s39064 bugnameltc-204586 severity-medium
targetmilestone-inin2404
--
You received this bug notification because you are a member of Desktop
Packages, which is subscribed to ubuntu-meta in Ubuntu.
https://bugs.launchpad.net/bugs/1833322
Title:
Consider removing
** Also affects: ubuntu-z-systems
Importance: Undecided
Status: New
** Changed in: ubuntu-z-systems
Assignee: (unassigned) => bugproxy (bugproxy)
** Tags added: reverse-proxy-bugzilla
--
You received this bug notification because you are a member of Desktop
Packages, which is
Pings done, in a perfect world (if all reply) that would cover more than
we ever need, but then there is 0% guarantee they even have time or care
about this at the moment :-)
If anyone has connections as well, please ask them to participate too.
--
You received this bug notification because you
Hi Dough
> If irqbalance is to be included by default, then there should be due
> diligence to demonstrate a clear benefit.
You are right that we should have that as well.
But this would be even more ture if this would be about "making it the default
when it was not before".
Right now (purely
I want to try to avoid that this becomes too stale, so I wondered
what we can do from here. Two things came to my mind.
On one hand I will try to use some indirect relations to pull in some
HW manufacturer experts. They often have large performance teams tracking
things like that against
Hi Paride
> Back in the day I asked upstream their take on irqbalance usefulness with
> newer kernels, here is their reply:
> https://github.com/Irqbalance/irqbalance/issues/151
Thanks for this and the other extra pointers.
The Debian bug was referenced before, AFAIC it is mostly around
a) the
Hi Mike
> SUSE ... says that the first step to get there is to disable
irqbalance
I've read the same, IMHO that is just "if you want to manually tune, disable
it" which does not imply that it is bad to have it. But this is how I read
it, I have not talked to the authors to get their underlaying
Hi Ethanay
> All I can find is a recommendation not to use it on CPUs with 2 or fewer
> cores as the overhead is said to be too high
This isn't a real problem anyway, the service will stop immediately if only
running on one core - even if running on multiple cores with the same
cache (as the
Hi Steve,
> I see a lot of strong opinions ... I would want any decision to remove
> irqbalance from the desktop to be based on evidence, not conjecture.
I agree that there is plenty of opinion (often backing up each other with cyclic
links) and not much data. Hence my compilation of the history
Hi, adding a couple of extra pointers here (I'm the Debian irqbalance
maintainer). This the Debian bug where the discussion on removing
irqbalance from the kernel Recommends happened:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=926967
In Debian irqbalance is not installed anymore by
Lots of good comments. I sort of agree with:
> So if we're going to make a change, there
> should be due diligence to demonstrate a
> benefit, it should not be based on
> Internet hype.
However, I would have said:
If irqbalance is to be included by default, then there should be due diligence
I said my initial piece and recommendation here:
https://bugs.launchpad.net/ubuntu/+source/irqbalance/+bug/2046470/comments/2
It carries through here... This was brought up as a recommendation in
Launchpad (here in this bug report) back in 2019, In that bug report, I
questioned why this had been
Status changed to 'Confirmed' because the bug affects multiple users.
** Changed in: irqbalance (Ubuntu)
Status: New => Confirmed
--
You received this bug notification because you are a member of Desktop
Packages, which is subscribed to ubuntu-meta in Ubuntu.
Does it seem correct to say that the general intention of irqbalance wrt to
system performance is to improve throughput (translating in some cases to a
more responsive system) at a cost of increased processing latency?
If so, then it should be considered and tuned generally with regards to
usage
Hi Christian,
I see a lot of strong opinions being given, but aside from the "don't
use it in KVM" guidance which appears to be based on GCE's engineering
expertise, very little evidence that irqbalance is actually a problem.
I think it's true that in the default config, irqbalance can interfere
Thank you for your incredibly thorough analysis of this. Since finding
this via bug 2046470, I have tried, without success, to create a test to
show any difference in performance or power or whatever between
irqbalance enabled/disabled on my Ubuntu 20.04 test server.
While my vote carries little
After all the history I was looking at where we are right now:
- irqbalance already is not in ubuntu-cloud-minimal images
- irqbalance is in normal cloud images and installed systems via the dep from
ubuntu-server
--
You received this bug notification because you are a member of Desktop
I subscribed a few people directly to get their input.
@Steve
I've subscribed you after trying to find, refer and summarize all of the past
to allow you and anyone else to read into this in one go. I think I'll need
your input as Architect and as participant of these discussions right from when
# 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
# Actions by Others
Times have changes, as mentioned above the kernel learned many new tricks.
More new I/O hardware virtual or physical appeared that tries to be smart
and thereby sometimes conflict with what irqbalance does.
Some are mostly based on the links referred above, the Debian
# Integration and maintenance
Despite some saying it is for the past only, it is regularly updated
and has multiple releases per year throughout all the time [4]. Those
updates flow well into Debian and Ubuntu - so it is not a classic "old
and outdated" case. And while not much changes in those
# Referred Arguments
An argument that might not have been so strong more than a decade ago
but is much more today is power savings and that is an aspect that comes up
over and over.
It also had reports of conflicts with power saving [10] and e.g. dynamically
disabling/enabling cores which is much
Hi,
this was overlooked for too long but came up in bug 2046470 again which made me
see this for the first time.
I'd wish we'd have had that even a bit earlier e.g. to release it with
mantic and not half way through noble, but still now is the time to
still change the next LTS.
I needed to make
I am using Ubuntu 20.04.1 LTS 64 bit on an Intel mobile CPU and Gnome
3.36.8 (Kernel 5.4.0-58-generic).
irqbalance is still installed by default.
The frequently used Gnome Extension "cpufreq" shows a permanent warning
that irqbalance is active.
If I uninstall irqbalance the warning is gone.
32 matches
Mail list logo
