All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
OpenVZ plans: What is considered abusive behavior and can get you kicked out? (excluding illegal)
What kind of activity in an OpenVZ plan can get you kicked out by the provider for ToS abuse? (beyond the obvious illegal activities such as spamming, DoS, etc)
What can get you kicked out: Can 100% use of each of CPU, RAM, SWAP, disk, network by one single container get you kicked off due to something like ToS abuse?
What is the effect on other containers? how to detect it?: If one OpenVZ container is doing this 100% scenario, how much effect is felt by other containers? and how do the other containers detect this from within their own container?
i've heard "fair share" being mentioned in plans. But is that fair share automatically enforced by OpenVZ/Operating system, so the container owner doesnt have to worry about restraining one's process? Or is it expected that the container owner perform some sort of restraint?
This question is not about how to abuse, but about how to not abuse but still maximize the value of what you have paid for.
Comments
What is classified as abuse is generally different from provider to provider.
If you are using 100% of each resource as you mentioned above and you did not specifically pay for dedicated resources that would be considered abuse.
The effect on other containers differs depending on what is being abused. Some hurts more than others but yes it can affect other containers. Most providers will allow you to burst to a higher load for a period of time but once you have past that point it is considered abuse. The only real way to detect it from another container is general slowness or I/O wait / high load in your VM with your server not being the cause of that load.
Fair share is not automatically enforced by OpenVZ. You should worry about restraining your processes. It can be limited via extra settings but most places to not enable them.
Hope that helps
If you think it might be abusive, it probably is.
If you want to use 100% of anything, it's probably abusive.
Aim always to make your scripts and services have the minimal impact on your own VPS. By doing that you probably won't affect other people. Look at 'nice' and (on KVM/Xen) 'ionice'. Look at bandwidth limiting for any app that transfers a lot of data. By limiting bandwidth you also limit disk I/O. There are many things you can do to be nice
Thx for the info. I didnt know about nice and ionice - will definitely look into these.
Its not that I don't know how to restrict my processes.
I am confused why LEB offers specify number of cores, when the real determining factor of how many median or max number of containers is never disclosed. Does this mean "fair at 7 PM at night" might not be fair anymore and considered abusive at say - 10 AM when other containers are more active?
It comes down to common sense, but some providers do list their limits:
https://bandwagonhost.com/knowledgebase/6/TOS---Terms-of-Service.HTML
https://crissic.net/transparency
Most providers doing this long enough do have a fractional value per core in either their ToS or AUP, along with burst duration. Multiply that with the number or cores in your package to determine the actual load factor.
If you are going to be using all the resources get a dedicated server.
We allow using all resources even on OVerZold servers, however, loads over the number of cores are only allowed on the Biz plans and no longer than a few days.
What gets you offline or even kicked is DDoS (second time without a DDoS protected service), outgoing DoS or spam (no matter if due to hacking or not), blatant service abuse (loads of tens or hundreds of times higher than your number of cores), no reply to tickets signalling you the abuse and no action taken (if we keep getting the same alarms about your service 24 hours later), things like those.
Note that most of the time is only suspension (except the attacks and spam that usually lead to direct indefinite suspension-until the service runs out) and you can resume service upon promise that wont happen again.
Disk abuse, continuous access for more than 10 MB/s usually is considered abuse if it lasts more than a couple of hours (and a sign of malfunction, such as swap used instead of ram due to memory leaks, for example).
Network abus, the mentioned outgoing attacks, sustained 10k+ pps, but we usually allow it up to 40 k if the service is legit, over 40 k bursts for longer than a few minutes, 300 mbps sustained throughput for long time. You usually get a warning about those and we also check it for forbidden software, such as card sharing for coded tv programs, tor exits, hacking and the like. No legit service needs more than 40 k pps to my knowledge, at least none that is expected to be run on a VPS.
Yes, I am serious, we do allow using all resources, only locking cores making them unavailable over OVZ is not allowed. There are some softwares that do that, especially mining.
Thx. I actually have few dedicated machines for my heavy loads.
My VPS are currently idle. The reason why I ask is it helps me decide whether I should just not renew the VPS or if I can figure out what load is considered reasonable to assign and then keep the VPS.
When I run "htop" - the CPU - I assume that is not "my share" of the CPU, but the "overall" CPU server usage right? RAM I believe is only my share?
Also, any other helpful tools besides nice and ionice?
To answer this question, it's important to remember what VPS actually is. Generally, a VPS is a combination of dedicated resources and shared resources. Typically:
Dedicated: RAM, Disk Space, and Bandwidth transfer.
Shared: CPU, Disk I/O, and Network port(s).
So to answer your question you can ask yourself: "Am I using 100% of a shared resource or a dedicated resource?".
Generally, if you are maxing out a dedicated resource, that is fine. However, If you are maxing out a shared resource, that is not fine. It's just like you think it would be. When in doubt, ask your provider. If you want a setup where you can max out all resources, a VPS probably isn't for you. Good luck
Unless you are running vServer or some sort of shell, the load is only your share. If your VPS is idle and you still see something there, that is a problem, you should see 0 all over unless really-really small vps and htop can generate enough load to be noticed.
Using munin, sometimes it's using about 20K packet per minutes
hi, my vps got suspend a while back with "bandwidth abuse"..
my bandwidth usaged is 20GB in 5 days, is this not normal?? i am using it for vpn.. and my vps bandwidth 500GB,. am i not allowed to used all my 500GB bandwidth??
and if my speed is 4MB/s for 1 hour long, is that categorize as network abuse?
No, that is not normal, but I also got suspended once for the same thing while feathur was showing normal usage, with the explanation it was not displaying correctly, so maybe was your case too.
Do you have any good offer at current time? :-D