Qubes slow (out of mem)

hey,
got some troubles.
today i received new ram and new nvme ssd, i inserted it in my laptop and everythings fine (seem so).
ok i reinstalled qubes,… if im running ~3 qubes xentop is showing me its using 30gb ram
kk, anyhow i got 64 probably these 30gb is statically, so i read this anywhere that xen is using ram statically, but ok was also wondering what happens with the rest of the 64gb? it wouldnt get used?
anyhow, i tried to install everything to get everything working for me

now i got the problem that all my qubes are using max mem (4gb)
doesnt depends if its my social media vm where discord, element, signal is opened
or my surfing vm with just only brave and one download opened.

if im running out of memory, dom0 is still good, i could use the menu and everything but all my qubes are stopping to work

does anyone got an idea what i can do here? i ordered 64gb to get some fun on qubes and to be able to do here my daily jobs (more than just surfing and social media as before with 16gb)

btw. looked up - my sys whonix is also using 4gb ram (the service)
vault without internet connection also needs 4gb ram for just keepass.

hopefully anyone can help me soon

Nice to letting us know, but what have you done to overcome this? This has been discussed numerous times, just search the forum. Experiment with balancing and manually alocating RAM to each qube as well as to dom0. Use zram tools. Etc…

this was my question, how to overcome this? thats the reason why i created this topic
this isnt a manual.
i also wrote “hopefully anyone can help me soon”

oh really? i didnt found topics about that.
memory balancing is enabled on every qube.

my question would also be, could it be that the new nvme ssd is getting hoter than the stock one? and thats the reason why my qubes is getting slower / unusable?
but this wouldnt make sense why dom0 is still working well

Qubes out of memory

www.catb.org/~esr/faqs/smart-questions.html

ok i scrolled back to october 16, idk where you saw the same issue as i have
i found only 1 without a solution

again:
all qubes are using max mem, also qubes like vault where i only got my keepass running

I’ll repeat myself

idk what you would mean with experiment with balancing.
i understand what you would mean with manual alocating ram but how? as i said, my vms are using 4gb ram and hang up if max is reached, i already tried to give my vms 6gb ram but it cant be, that my vault is using 6gb for only keepass
this isnt realistic.

why should i alocate ram to dom0 manually if dom0 is still fine and not affected?

i dont know tools like zram

but if you dont help me further, why are you replying to my thread and just bully around here?

go away from this forum if you got a bad day…

1 Like

When this happens, the worst thing you can do is whine about the experience, claim to have been verbally assaulted, demand apologies, scream, hold your breath, threaten lawsuits, complain to people’s employers, leave the toilet seat up, etc. Instead, here’s what you do:

Get over it. It’s normal. In fact, it’s healthy and appropriate.

We’ve found by experience that people who are careless and sloppy writers are usually also careless and sloppy at thinking and coding (often enough to bet on, anyway). Answering questions for careless and sloppy thinkers is not rewarding; we’d rather spend our time elsewhere.

So expressing your question clearly and well is important. If you can’t be bothered to do that, we can’t be bothered to pay attention. Spend the extra effort to polish your language. It doesn’t have to be stiff or formal — in fact, hacker culture values informal, slangy and humorous language used with precision. But it has to be precise; there has to be some indication that you’re thinking and paying attention.

Spell, punctuate, and capitalize correctly.

Be precise and informative about your problem

Describe the symptoms of your problem or bug carefully and clearly.

Describe the environment in which it occurs (machine, OS, application, whatever). Provide your vendor’s distribution and release level (e.g.: “Fedora Core 7”, “Slackware 9.1”, etc.).

Describe the research you did to try and understand the problem before you asked the question.

Describe the diagnostic steps you took to try and pin down the problem yourself before you asked the question.

Describe any possibly relevant recent changes in your computer or software configuration.

If at all possible, provide a way to reproduce the problem in a controlled environment.

Describe your problem’s symptoms in chronological order

Describe the goal, not the step
Often, people who need technical help have a high-level goal in mind and get stuck on what they think is one particular path towards the goal. They come for help with the step, but don’t realize that the path is wrong. It can take substantial effort to get past this.

Stupid:

How do I get the color-picker on the FooDraw program to take a hexadecimal RGB value?

Smart:

I’m trying to replace the color table on an image with values of my choosing. Right now the only way I can see to do this is by editing each table slot, but I can’t get FooDraw’s color picker to take a hexadecimal RGB value.

Be courteous. Use “Please” and “Thanks for your attention” or “Thanks for your consideration”. Make it clear you appreciate the time people spend helping you for free

Don’t flag your question as “Urgent”, even if it is for you

How To Interpret Answers

RTFM and STFW: How To Tell You’ve Seriously Screwed Up

There is an ancient and hallowed tradition: if you get a reply that reads “RTFM”, the person who sent it thinks you should have Read The Fucking Manual. He or she is almost certainly right. Go read it.

RTFM has a younger relative. If you get a reply that reads “STFW”, the person who sent it thinks you should have Searched The Fucking Web. He or she is almost certainly right. Go search it. (The milder version of this is when you are told “Google is your friend!”)

In Web forums, you may also be told to search the forum archives. In fact, someone may even be so kind as to provide a pointer to the previous thread where this problem was solved. But do not rely on this consideration; do your archive-searching before asking.

Often, the person telling you to do a search has the manual or the web page with the information you need open, and is looking at it as he or she types. These replies mean that the responder thinks (a) the information you need is easy to find, and (b) you will learn more if you seek out the information than if you have it spoon-fed to you.

You shouldn’t be offended by this; by hacker standards, your respondent is showing you a rough kind of respect simply by not ignoring you. You should instead be thankful for this grandmotherly kindness.

If you don’t understand…

If you don’t understand the answer, do not immediately bounce back a demand for clarification. Use the same tools that you used to try and answer your original question (manuals, FAQs, the Web, skilled friends) to understand the answer. Then, if you still need to ask for clarification, exhibit what you have learned.

For example, suppose I tell you: “It sounds like you’ve got a stuck zentry; you’ll need to clear it.” Then: here’s a bad followup question: “What’s a zentry?” Here’s a good followup question: “OK, I read the man page and zentries are only mentioned under the -z and -p switches. Neither of them says anything about clearing zentries. Is it one of these or am I missing something here?”

Dealing with rudeness

Much of what looks like rudeness in hacker circles is not intended to give offense. Rather, it’s the product of the direct, cut-through-the-bullshit communications style that is natural to people who are more concerned about solving problems than making others feel warm and fuzzy.

When you perceive rudeness, try to react calmly. If someone is really acting out, it is very likely a senior person on the list or newsgroup or forum will call him or her on it. If that doesn’t happen and you lose your temper, it is likely that the person you lose it at was behaving within the hacker community’s norms and you will be considered at fault. This will hurt your chances of getting the information or help you want.

On the other hand, you will occasionally run across rudeness and posturing that is quite gratuitous. The flip-side of the above is that it is acceptable form to slam real offenders quite hard, dissecting their misbehavior with a sharp verbal scalpel. Be very, very sure of your ground before you try this, however. The line between correcting an incivility and starting a pointless flamewar is thin enough that hackers themselves not infrequently blunder across it; if you are a newbie or an outsider, your chances of avoiding such a blunder are low. If you’re after information rather than entertainment, it’s better to keep your fingers off the keyboard than to risk this.

Questions Not To Ask

Here are some classic stupid questions, and what hackers are thinking when they don’t answer them.

Q: Where can I find program or resource X?

Q: How can I use X to do Y?

Q: How can I configure my shell prompt?

Q: Can I convert an AcmeCorp document into a TeX file using the Bass-o-matic file converter?

Q: My {program, configuration, SQL statement} doesn’t work

Q: I’m having problems with my Windows machine. Can you help?

Q: My program doesn’t work. I think system facility X is broken.

Q: I’m having problems installing Linux or X. Can you help?

Q: How can I crack root/steal channel-ops privileges/read someone’s e-mail?

1 Like

ok, i dont want to answer fully to your post.
i agree that my grammar wouldnt be the best, but im not native, so sorry.
i agree, that i didnt punctuate and capitalize correctly. would punctuate if this is so important for you but you dont do it. so you cant desire something you dont do.

i were precise, but you arent reading meaningfully. we saw it as you send me the qube search link to my topic.
there is only 1 topic which is similiar to mine but without solution. so it seems like you dont want to help.
ok, thats a thing most people had to learn.

i never flagged my question as urgent, you also arent able to do that in this forum. my topic is pretty neutral and my text was just something like: i would be glad if i could fix this soon.

English is not my first language too. I tend to edit my posts until they’re grammatically and semantically correct, according to my knowledge.

Disable balancing and check the forum how much each kind/type of qube would need RAM. I have 16GB RAM and am running 12-18 qubes, including multimedia and Windows HVM.

i also edited my posts till its correct to my knowledge.

disable balancing would / could fix this? could try this out but im unsure because keepass is freezing also with 4gb ram. since the last reboot everything is running fluent but still using max mem. could maybe disable balancing and downsize my qubes.

oh you are really using only 16gb ram? but these 12-18 qubes arent running at the same time right?
because as i had 16gb ram i ran out of memory (qubes didnt start anymore qubes - so this common message show up like “not enough memory to start qube”.
and i only had opened my daily driver qubes without my “job” qube where i would code,…
also got sometimes the problem that i wasnt able to watch youtube with my 16gb ram.
had to close some tabs that my qube was running well again.
maybe is brave not qubes optimized and thats my problem?

brave is my standard browser because i got the feeling brave is the best privacy browser out there atm.

That’s good and appreciated. Capitalize “i” as well. It’s you!

Are.

It would freeze with 64GB as well.

My vault and Keepass in it is running on 400MB.

I told you everything. Now you search and experiment.

So, this topic is about supporting your confidence or about “out of mem”?

… and get back with findings on experiments to help others too.

thats what i said, we are in the internet, wouldnt see the sense to capitalize things here.

yeah, thats the reason why i was wondering why disabling balancing should fix my issue?
and why it should run “fluent” with 400mb. but as i said, i would try it out. thanks for this tip.

sorry, i dont get this answer, tried to translate it with something similiar like google translate (just from startpage) because i only know confidence in a different context.
but to give a answer anyhow:
i dont getting anymore (since i upgraded to 64gb) these “out of mem” messages from qube
so i guess you mean with “out of mem” these “cant start qube” messages.

Here we go again.

whats your problem?
i am not the only person here who isnt capitalizing things and i know you are the only person who is ONLY saying to me that i have capitalize things
i dont want to attack someone but why didnt you told newbie he should capitalize things?
you wrote with him on the same topic here:

clodius is also not capitalizing:

here another one:

and you are still the only person who want that i only capitalize things. in these posts i posted here, nobody got a problem with it that they didnt capitalize theyr words.
also all my previous posts arent capitalized. who cares? if you are googling for a problem it doesnt care you if this post is capitalized or not, the only thing what would be important is, that it would help you further.
but i dont want to continue this discussion, thought we should keep this forum clean.
so everyone who is searching for a topic similiar to mine doesnt have to scroll down over ~100 senseless posts.

Your posts. The answer was in post #2. Any caring user will find answers there.

And contributing things can be found exclusively in my posts. All of them. Not in a single one of yours.

oh you think so? i thought the goal were that everyone could use qubes.
sorry, i misunderstood this then.
asking me then, why qubes should be good for persons who are journalists and never touched linux in any way.
and why qubes is advertising that qubes is good for journalists who are working on sensitive data.

there are 2 ways you are reacting as you react:
1.) you want to be a wannabe senior who forgot the goal of qubes.
2.) you got some complex which you cant process and have to let it out on other people.
maybe you didnt reached a goal or something like that and thats the reason you are acting like someone who act on you like that.

then tell me what were contributing of your posts?
ok maybe only the thing you told me i could disable memory balancing, the rest of it were how i have to write, how i have to ask things,…
thats not the topic of my topic.

and yeah, sure, you never found something contributing on my posts because im a planless dude who was wanted to get some help.
but because you are counting your “tutorial how to ask questions” also as contributing, then my answers should be also a contribute.
because thats my opinion.
i asked my question correctly, i explained what my problem is,… everything should be fine.
but i cant be responsible for that, that you dont read my post meaningful.

this would be my last post.
maybe a forum mod would close this topic and open up a new one for me or something like that without your answers which are so super contributing to the non techie qubes community.

Gentlemen.

I do a lot of errors when typing. I edit my posts constantly.

But I never get into fights in forum. I decide to pass on and not answer to things I can’t. And breath between things misunderstood and meditate on content of my last post when people ask questions over I already answered. Maybe I shouldn’t have answered if I was not clear enough. Maybe I was not in the right state of mind when I answered. And I learn from that. If the question is too vague, I try to teach without being hurtful, without judging. If I’m writing and I judge, I dismiss my post and I say to myself that someone else will answer, unless I’m tagged. If I’m tagged I answer only when I’m available. We are all here to learn and exchange…


Qubes uses basic , simple, home-made memory ballooning. When a qube is created, its either a hvm with static memory assignment or a pvh qube with dynamic memory assignment, defining max and min memory settings.

When a user starts a qube with dynamic memory and memory balancing enabled, a qube will start with all the memory it needs, including a lot of virtual memory which isn’t really needed. That memory will stay in the qube unless another qube needs to “steal” that memory, which will trigger dom0 memory ballooning which will try to steal unused virtual memory from one to give it to another qube. Unfortunately I do not know all the details here, but the result of this is that qubes will show less memory usage in active qubes widget over time.

What you seem to observe is two separate things: you are one of the lucky one having 64gb of ram, which means all qubes can use their maximum dynamic memory settings at the same time. Until all 64gb are used across qubes and dom0. At that point, and when that happens, slowness could happen, but should not happen prior of that point unless all applications are actively used (cpu actually trying to do something in that allocated memory).

So to separate things:

  • having your qubes use maximum dynamic memory setting is normal until physical memory constraints are met.
  • crashing or slowness should not happen until those 64gb of ram are physically used + swap space
  • somewhere in the middle of this is those 4gb of ram actually being reserved and used, with slowness happening which involves in-vm memory being swapped to qube’s swap partition and or dom0’s swap switching in-memory with disk swap because too much memory is actually being physically used. From what you describe this is not the case.

Now. Other threads will talk about current issues with performance. On my side, everything updated, I experience critical slowness as opposed to months ago on my x230 when one qube is actually using all it’s ram (8gb) and hugging its two assigned cpus, leaving little place for others to do their work, and where other qubes are balancing as they can (16gb physical memory here), reducing performance to the point dom0’s controlled mouse input stops reacting to my movements. That was not like that before where dom0 stayed responsive in that case. I have not yet isolated the issue but I know i’m not alone reading other threads and searching and posting where relevant.

Those changes if behavior are definitely not normal, but an understandable consequence of attempts fixing known issues from Qubes OS side and lack of actual testing of testing packages before they land into stable repositories, which I am partly responsible because I do not always test and report bugs on testing packages as many other advanced users, by lack of time testing those packages on my daily used laptop. There are separate issues you might have missed.

Question here is which model of hardware and describing hardware model, cpu and difference of behavior since approximately when.

In other terms: Was it always like that on that particular piece of hardware? Is it happening under specific workload happening under certain templates based qubes?

I see this post having quickly slide into nonconstructive posts and i’m sorry about that. On the other hand, I know nothing about your machine outside of a memory upgrade you recently did. What would be helpful is to know what were those memory chips. Were they optimal for your hardware? Are they the cause of your now experienced slowness as opposed to before?

Otherwise this post, as of now can be read as: “Qubes OS is slow on a 64gb machine”. Unfortunately, there is nothing actionable on that statement, where I agree second post of this thread said the same things I said previously, but in different words, on memory ballooning of qubes and telling you this is quite normal. But slowness is not.

So more detail on what was used before, and now would help. Machine details (hcl is better) would help everyone trying to help you back.

I was trying here. Sorry if I missed details trying to skim nonconstructive replies in this thread. If important details were missed, sorry but not sorry.

Please restate them (quite relevant parts) and try to not take language differences personal. I’m also not native english. And i’m not sorry about this. And I also not capitalize all my “I”. My goal is to help. Not to be always perfect. But everyone is different and I don’t care. Aho.


Short version:

  • Title of this issue is misleading considering provided information. You might want to rename it since nothing points to " out of mem"
  • You seem to experience slowness. But what is the point of reference on that same hardware? Share a hcl report maybe?
2 Likes

hey,
thanks for spreading love over this forum <3
im also more team love in the internet, knowledge should be available for everyone. there are 2 things why people arent asking precisely:
1.) he really doesnt get that knowledge
2.) people didnt searched in the web for theyr question
2. point ive done, didnt found my problem really.
and i also agree, if the other person is still asking although you answered the question, its possible that your answer wasnt that precise or the other person didnt got that knowledge.
i worked in a callcenter, i also had to learn answering people (also non techis)
everyone (especially in the internet) is on a different knowledge level.
and i guess our (or qubes) vision is, to be able to offer qubes to everyone who wants to life in a free and privacy friendly vicinity.
again:
thanks for spreading some love <3

yeah, im using pvh mode exclusively besides of hvm qubes but i didnt got some since this laptop is running on a fresh setup.

no problem that you dont know the details about ballooning.
but thanks for this “detailed” explanation (for me detailed, probably for an advanced user you could explain much more, but this makes sense for me, thought this is a error because with my 16gb ram all my qubes were scaled)

i read this point ~5 times but im unsure if i got this.
if my qube is using more than 4gb (maxmem) ram, it would write into my dom0 swap?
didnt checked my dom0 swap, but as you said, wouldnt believe that this was the case because if so, then dom0 should be also slower?

laptop: lenovo p15v
cpu: intel i7-10850
ram now: https://www.amazon.de/gp/product/B07ZLCVKPV/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1
stock ram: Lenovo ARBEITSSPEICHER 16G DDR4 3200 SODIMM Samsung

behavior is since i swapped my ram and my nvme. reinstalled qubes and yeah, tried to setup the new qubes os as the one before were.
since the last vm freeze (shortly after the installation) i didnt had it, that was also my question, if it could be, that my new nvme ssd is getting hotter than the one before?
wouldnt assume that because the one before also got a thermal bad - if that was enough for the nvme before, it should be also enough for this one?
because if the nvme is getting to hot it would slow down and at some point would stop working afaik.
but this wouldnt make sense again for me because dom0 were responsive.
if it would be a thermal problem dom0 should be also a bit slower right?

with the 16gb ram i never expected this.
if its happening under specific workload, idk. i had this issue 2x after the new installation and since then, everything is ok again.
lets see, ive didnt done that much anymore since the second time.
lets see if i could figure something like that out later (tomorrow)

maybe i find something what could stress my nvme.

but i wouldnt assume its about special qubes.

oh oh. dont say nonconstructive lol
in enmus opinion it was contributing lol
but you dont have to be sorry. every community got some… strange people. :slight_smile:
i wouldnt take that personal. i thought i knew qubes vision and people like you are confirming the vision of qubes. :slight_smile:

you should got now my setup, lenovo p15v gen1, cpu info and i also send you the amazon ram link.
anyhow, i would post here also the hcl as you mentioned it anywhere below.

thanks again <3
didnt know what you would need for a better help… i mean yeah ok to be fair, to post the hardware were also bit logic, but didnt thought about it, so sorry.

this should be a knowledge base and no wordbook.

sorry to ask, did you got a better topic title idea? tried to give my best but, since im trudge in the dark, idk how to call it better.

guess we wrote a while ago but never took a look at your profile - are you the owner of insurgo? thought you just got the name of the company o.O
if you would be the owner of it, i would have respect to you, i mean, you must be pretty busy with a business nearby.
anyhow, would be thankful to everyone who would help me. :slight_smile:

Qubes-HCL-LENOVO-20TRS1QE00-20221129-213457.yml (846 Bytes)