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
Why stop at 85% discount? Ask the OP to pay you 85% for using each VM.
i just request Sir..
The drives are slowing down in some nodes with older NVMe (as you've seen with a couple of benchmarks in this thread) so they are scheduled to be replaced to improve performance and to prevent guest KVMs from running on degraded RAID arrays. There are no errors on these drives, this is all preventative, and the maintenance will be completed with no downtime for anyone.
Looking forward to have SG general available
@MaxKVM do you allow VPS transfer
Transfers are not permitted.
In other words, you wouldn't tell users how much CPU they are buying. Instead, you can arbitrarily define a potentially very low limit (e.g. 5%), and send warning to scare a user who paid annually. Combined with not permitting transfers, the user who received such warning is forced to either idle their VPS for the remainder of the year, or pay an expensive price for a dedicated core although they probably only needs 30% or 50% core. NOT FRAGRANT!
For reference, Virmach has a published guideline of 33% of allocated cores, in that they promise to not send warning if usage is below this amount, although the actual limit being enforced is higher. It's reasonable to conceal the enforced limit as it could change over time, but it's also necessary to publish a safe level so that users do not feel a suspension hammer hanging over their head.
In fairness to @yoursunny, I agree wholly with his argument.
On the other hand, I have an account with another provider with (almost) the same policy; as long as my CPU usage doesn't endanger the overall health of the node, I'm fine to use as much CPU as my allocation permits (until I get a ticket
).
How do you replace SSD without any downtime? I'm interested in the technical. Do you replace some disk on the raid first then add more later after the first synced
That would be an interesting process, but we just live migrate all KVMs to other host servers before the system is taken offline for maintenance.
thanks for the explanation @MaxKVM
Although I agree with your sentiment, I have to admit the deals are still fragrant. I'm sure it's not a situation where they set a low arbitrary limit, but possibly a situation where they either don't want people to utilize the limit 24/7, or maybe they don't know what's a reasonable limit themselves. People would be mad if they gave out a number, and that number had to go down in the future because servers get filled up.
I could be wrong, but I imagine their LA and SG nodes are very packed compared to their other locations. So it's likely the cpu policy could differ between locations too. But admittedly, I am a bit concerned they are not getting more nodes but simply adding more RAM/NVMe... Wait it out a few months and we will know if the deals are fragrant. For now, it's looking good
Prem and fragrant..
Was going to pick up 2 more SKVM-4G-4CPU before 12 midnight PST time. Looks like the offer closed early.
You still can purchase any package at 50% off, coupon: edu50
Only for .edu addresses? Anyway if not I plan to add-on my existing ones.
Can I use this coupon many times buying different VM, or each account is allowed to use "edu50" for only one time?
How soon can I buy Singapore VM on your website?
I don't know. test yourself
MaxKVM said, Singapore location will be available before 01-31-2021
Virmach gave the number 33%. It means that you can safely utilize up to 33% without worrying about the suspension hammer.
It does not mean you would be suspended as soon as you exceed 33% - that's the (hidden) enforcement limit, which would be higher than 33% but can change over time.
It also does not mean that the server would have the capacity to allow everyone using 33% at all times - that's what allows the provider to oversell, because not everyone's workload is CPU-bound.
Importantly, Virmach promised not to suspend anyone if their usage does not exceed 33%.
If enough users on a particular host node are utilizing CPU so that the host node has insufficient CPU power, but nobody is exceeding 33%, Virmach would increase the capacity by migrating some users to other host nodes.
If someone is exceeding 33%, Virmach can suspend them.
Of course Virmach's CPU is much worse than MaxKVM's EPYC, and 33% of a Virmach core makes the safely usable CPU even slower.
Nevertheless, when I'm using / developing experimental software that usually consumes 3% CPU but might have 100% CPU spikes when I'm not looking (e.g. a malformed packet triggers a bug causing an endless loop, which had happened in the past), I want to have the peace of mind that the software would not trigger the suspension hammer.
By knowing a safe limit, even if it's lower than what I'm actually allowed to use (i.e. the enforcement limit), I can configure that limit into systemd or Docker, to guarantee that the software would never use more than the safe limit.
@yoursunny Why don't submit a ticket and ask. I will do myself too now, I guess 30%
You can certainly use 33% sustained without interruption. Nothing will stop you from using that little vCPU.
There are services running at all times that stop the worst offenders, and for the most part the only thing that happens to them is temporary CPU throttling to the extent that the performance of each core is (at worst) comparable to an 8 year old Xeon core (for a short period of time). Generally speaking, if steal is above 0-1% then you may have been abusing CPU. Contact support for logs and details. If support contacts you first then you may have been running too many benchmarks. 
Great. This answer would make MaxKVM fragrant again.
As long as it's not a suspension or a warning email that threatens a suspension, it's fine.
I've been terrorized by Virmach hammer and thought every provider has the same hammer.
I don't run many benchmarks. I just wanna encode push-ups.
ffmpeg is CPU-bound workload and could run for hours, but I can easily set 30% limit on the process. It would be faster than the crummy Oracle Cloud (nominally 1/8 vCPU, no suspension hammer but huge steal once exceeded).
ffmpeg is a monster.
I once asked buyvm support how to precisely set cpulimit value on my box if I dont want to cross the line. The reply was they already had solution for this (eqipment upgrade or sth?) and I felt less worried when running ffmpeg (avideo encoder). Of course I only do with those short mp4s so it won't last long.
Yeah its monster, like Plex, transcoder is based on FFMPEG. Not recommended on shared resources like vps
@yoursunny I would not believe in assurances from Virmach
If I'm approved for 33%, I'll generally set CPU limit to 25%, leaving some CPU for SSH file transfer.
I'm out of push-up videos though. Chest too sore.
I tested some other workloads in Docker, and it seems that the CPU limit is quite precise.
I'm gonna cancel one of the Virmach box next year.
They allow 33% CPU but also have limits on I/O, load, etc that is difficult to enforce precisely.
I know it's a bit off-topic, but does anyone have any experience with Windows on MaxKVM?
There are only three resolutions available: 800x600, 1024x768, 1280x1024, but I need more resolution options.
Does anyone have any solutions for this?
If you have desktop on Windows Server, you are doing it wrong.
mstsc.exe("Remote Desktop Connection") over RDP protocol. mstsc would resize the desktop to match client resolution.