New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
I was talking to a server at IHOP and she mentioned that the highest she has seen is "42 plates pancakes" consumed at all-you-can-eat-pancake promotion
I am just wondering how much money that customer's primary doctor made?
For the right price, I can be your doctor
i want
unmetered cores
unmetered bandwidth
unmetered storage
unmetered ram
unmetered ips
unmetered ddos protection
unmetered ticket
unmetered chicken
ofc budget : 2$/year
How about "unmetered CPU"?
unmetered response time
dedicated burst thx
With shared economy vcore, ok?
Unlimited drama threads on LET
bitcoin fixes this
Nope, we have a few hundred VDS running this workload per location. All hypervisors are healthy and performance is as expected.
Does anyone have any idea why github banned Quilibrium client?
https://github.com/QuilibriumNetwork/ceremonyclient
It's always the few who ruin everything for the many. Like yes, maybe you have a good argument about being sold dedicated resources and not being able to run them 24/7. But when you stand on that platform and push it that hard, the net result is not better for anyone than it was before you did. Everyone was happy with netcup. Crypto bros ruin everything.
To the point.
I still have hard time though explaining why the mob is attacking to the providers instead of the miners.
Because attacking the miners is a 150% futile exercise. These are - like it or not - problems the providers have to solve. Also there more than just a purely idealistic point fingers at the bad guys level. Blame is irrelevant for determining that calling something dedicated which really isn't is giving people a wrong impression of what they are buying.
"But, but, but, ... the calculation would work if it weren't for those evil miners!" is not a valid argument. This is the real world. The real world contains miners and there is no joker card that says "as long as the evil miners are to blame you are not responsible for providing what you sold to your customers". Solutions need to be found and proclaiming how unfair this is is not a solution.
The easiest solution would be to simply not make grand claims that can't be realized (aka just being honest about the product) or making sure that they can be by enforcably excluding the miners (well really any kind of user planning to run CPU intensive applications for extended amount of time) and maybe making an effort to really isolate the resources available to each user.
@jar is absolutely correct if this CPU mining craze continues it'll likely affect less extreme users of the products in question but the one interested in and (hopefully) able to correct the situation is the providers. As long as they are offering these guys a bargain they'll take it and them taking the provider up on the offer results in the overall product being degraded something is wrong and needs to change. This also makes claims along the lines of "No interference from neighboring VMs whatsoever" (which is actually claimed pretty much verbatim) obviously untrue, so they need to either be redacted or made true. If that means biting one or the other sour apple then that's too bad.
Yeah, i know, i'm pretty much wasting my time here as you'll just keep foaming at tHe miNeRs but like i've said that's not even going to make a dent of a difference.
I get the principle of it, I do. But the maximum number of happy people and the lowest financial impact on the customer spread across all of them is had by banning the use case. Because a reduction in service quality or an increase in price for everyone across the board is the only other answer, and even if the 1 person is morally correct that's hardly a comfort to everyone negatively impacted by it.
2 Questions:
1) How can you detect miners? (without making more Karens open threads about data privacy)
2)
These are the recommended settings for the quilibrium shitminer. I hope it's obvious that if these buffers are actually utilized all the time, the memory controller will be brought down to knees, no matter the VDS no matter the provider, no matter the virtualization.
Do you think that this is normal use that it's the fault of the provider that has not anticipated?
I fully agree. It's not like i feel the mining would be somehow worthy of protection. Like i've already said in the other thread, while feel that marketing one thing under the label of another is not the most transparent approach, it's still pretty much the same for most practical proposes as long as the final result comes reasonably close. Finding ways to identify and boot (or ideally not even take on in the first place) the miners after adding a short and friendly line to the AUP is pretty much the most pragmatic and straight forward (it clearly communicates that the product is not designed for this and people that want this should look for another - likely more expensive - provider, which is prepared for it) approach here.
Network traffic and keeping one's mouth shut about the actual procedure. Giving out hints how to evade detection would be stupid anyways. Obviously a constant uniform load is also pretty much a tell tale sign but then that's kinda the same for various other long lived high load tasks, so it would likely result in a couple false positives and i gues that since it's, as i've learned, just the evil miners who are ruining the party these guys should be left alone.
If that doesn't work out what's left are the other options (drop guarantees or increase reserves with the latter likely resulting in price increases).
... A virtualization environment that can be that easily perma-DoS'd by the guest might want go back to the drawing board ...
... Do you have a list of anticipated use cases for me? ...
We had this already in the other thread. It's anticipated that the client will use it's VM to run some software. If that breaks the provider's calculation it needs to fix it one way or another. Crying about buffer sizes or how unfair it is to run resource hogs will fix exactly nothing for any client whatsoever.
Ok let them alone, so
This is not an issue of the virtualized environment, this is how computers presently work. Not to mention the fact that the kind of shitload that is usually added by the miners is more likely to cause the whole CPU to throttle, as many people have already pointed out multiple times.
In fact it is the HW that does not allow perfect isolation, especially in such use cases, not the evil provider or the virtualized environment. And that joke I read multiple times about pinning, does not work here.
The best thing the provider can do is to mitigate the shitload to different nodes, until all the nodes are being into shit by mining.
As me and others have tried multiple times to explain, at the moment there is no way to isolate shit. It's not that the evil provider tries to fool you, it is that this is shitload, and it will cause problems unless the host runs a couple of VMs.
So do you think it's wise to leave the miners alone?
Sucks for them. Guess they can't give any guarantees then.
Well, not really as you are awfully generalizing/simplifying here. Also, as i've already told you before: Not being able to perfect something or the solution being unpleasant doesn't excuse not trying to begin with.
Did you actually read my post?
One could say that in a court it would be easy for the provider to prove that you get what you paid, a core. Now if EPYC throttles the cores when the whole CPU is doing shit, well go sue AMD...
...in practice though, no one of these German providers has this mindset. They mitigate load as long VMs get the load they actually use. Recent past proves that this works pretty well by how many people are happy with the performance of these machines.
So I am not sure I can agree with that "not trying".
Sorry, it seems I was confused by syntax.
That's based on how many actual samples of how many architectures? You realize that it's just basically not seeing an improvement for the HT threads or it might be bound by a power draw limit, which could - if existing - be seen as not exactly ideal? That could be something to further investigate? Nah, fuck that, with this massive data foundation it's 100% certain that any kind of improvement is impossible. Pointless. If there was a possibility you'd just again move the goal post to "But the memory can bottleneck" and therefore declare it futile.
Yeah, that suddenly makes a claim like "No interference from neighboring VMs whatsoever" true. I see.
While were at it: Do you maybe fancy buying a space shuttle? Well, OK, it's a paper plane but i can't improve it further and all my friends say it's the real thing, so it obviously is the real thing.
Which part here describes said attempt? Could it be that just doing what's convenient to oneself and ignoring everything else might be a little insincere?
Computer architectures, even on servers, are usually optimized for efficiency on specific workloads and not isolation.
You will see the same results even with HT off, and I am not sure it's thermal. My bet is yes.
Leaving shared components like memory controllers aside, I think that it's safe to say that the exception in performance is when one core runs alone, as it gets boosted, not the other way round.
In CPUs with complicated caches like 7950X3D, it's gets even more complicated to predict how performance scales with cores.
When someone is in computing business, one way or another, I think he or she has the common sense to understand what they are buying when buying a VDS. I found it extraordinary that so little people understand how these products work, "pinned" or not.
The expected performance from this product is of a dedicated core. If not you open a ticket and it gets fixed. Win-win.
So you are actually implying what, that there is no such thing as a VDS and those who sell it are liars?
My point is that a VDS is a product that history shows that it works pretty well when it is used for it's intended use. For the previous reasons, mining does not compile with this, and this is why it's not allowed.
At the end of the day, by how universe works, there is a way to abuse a VDS and it is mining. The problem is the product or the abuser?
What netcup could do in this case in your opinion now they are into shit? Or what could they have done? As far as I know they do not allow mining, but here we are.
Yeah, it's OK. You are right. I'm bored with all the circular arguments. I've just read the first couple of lines and it went straight into full on speculation territory. No thanks. I've made my points. If you don't like them that's too bad but these kind of scattershot discussions where one thing always leads to 5 dozen others are simply to pointless and annoying. With every single post you broaden the scope further...
I am sorry if world is too complicated for you.
I'll just let the uninvolved onlookers be the judge of that.
@TimboJones vs @jsg - I just can't tell who is who 🤷🏻♂️
Oh, I thought I've replied that.
There is no single mining pool, we are talking about random encrypted traffic, so there is no such magical thing to detect network traffic.
Also, you cannot ban constant uniform load as it's not against AUP.
Other than that, you have to prove that your client runs mining otherwise you will get a shitload of disputes and threads like these are not my dedicated cores.
Yeah, no. The healthy mindset is to attack miners instead of being part of the mob writing things out of my butt about how VDS should work.
I think that's probably what you'd have to do to impact the least number of users. Keep everything the same and kick out anyone who looks like they're running it based on resource usage patterns, then turn a blind eye to the complaints. Probably, honestly, wouldn't have many false positives if you truly gathered good data first.
@itsdeadjim alright, you know it all, seen it all and your definition is the only one that counts. If that gives you a warm and fuzzy feeling so be it.
By the way: That's probably also what those miners are thinking about being attacked. Some weird guy writing shit on a message board... well who the fuck cares? Colossal waste of time but suit yourself.