Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Shells Virtual Desktop
BMail.ag - Secure Email Service
Server.net
CPLicense.net
VPS Server
Buy VPN
Vultr
VMs for AI
HostDare
HostDare
ReliableSite White-Label Dedicated Hosting for Resellers
InterServer VPS
BMail.ag - Secure Email Service
Best VPN
High-Performance Bare Metal Server Solutions
Karvl.com
Server Mania Cloud Hosting
DataWagon Hosting
AlphaVPS Hosting
Evoxt.com
Clouvider
VPS Hosting with NVMe
Residential IPs in the US & 4G Mobile Proxies in EU & US with Unlimited Bandwidth
ReliableSite White-Label Dedicated Hosting for Resellers
Rabisu - Hosting Solutions
Shells Virtual Desktop
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.

Delux host cpu steal on vds plans?

edited June 1 in Help

Greetings
First of all I gotta say i'm not accusing @DeluxHost of anything. I had a couple of services with them and im quite satisfied, I just need some expert's opinion.
Im running a vpn service with arounf 900 users, im using xray core, my vps services were too weak to handle this amount of users so i purchased the first vds plan on their website(8g 4c). expecting to be able to handle a few more users as well but im getting random delays, high pings and random disconnections. I asked chatgpt(as any newbie person would) and i was asked to run a few commands and was told that

Your VPS is not underutilized — it is:

❌ heavily CPU stolen (~40%) → host oversubscription

That is why you get delay even at “50% CPU”.

root@Static:~# ss -s
sysctl net.netfilter.nf_conntrack_count
sysctl net.netfilter.nf_conntrack_max
top -bn1 | head -5
Total: 9200
TCP: 16913 (estab 8880, closed 7980, orphaned 39, timewait 7978)
Transport Total IP IPv6
RAW 1 0 1
UDP 78 2 76
TCP 8933 3546 5387
INET 9012 3548 5464
FRAG 0 0 0
net.netfilter.nf_conntrack_count = 34467
net.netfilter.nf_conntrack_max = 262144
top - 15:09:41 up 19:57, 2 users, load average: 2.70, 3.67, 3.74
Tasks: 169 total, 1 running, 167 sleeping, 0 stopped, 1 zombie
%Cpu(s): 10.8 us, 9.2 sy, 0.0 ni, 38.5 id, 1.5 wa, 0.0 hi, 6.2 si, 33.8 st
MiB Mem : 7894.1 total, 5727.4 free, 1446.6 used, 974.5 buff/cache
MiB Swap: 1024.0 total, 1024.0 free, 0.0 used. 6447.5 avail Mem
root@Static:~#
and
15:11:00 CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle 15:11:01 all 8.79 0.00 6.34 0.00 0.00 4.91 41.10 0.00 0.00 38.85 15:11:01 0 10.24 0.00 4.72 0.00 0.00 2.36 39.37 0.00 0.00 43.31 15:11:01 1 5.26 0.00 4.39 0.00 0.00 13.16 46.49 0.00 0.00 30.70 15:11:01 2 9.23 0.00 6.92 0.00 0.00 3.85 40.77 0.00 0.00 39.23
so here I am asking for some guidance on what's going on here

Comments

  • They reached out to me and fixed it.

    Thanked by 2zed forest
  • forestforest Member

    Did you open a ticket before opening this thread?

    Thanked by 1equalz
  • woinokizwoinokiz Member

    I think it's always better to contact host atleast once before coming into forums, as this will stay in records forever

  • CalypsoCalypso Member

    @forest said:
    Did you open a ticket before opening this thread?

    Like any newbie would do. And a sensible newbie would not use ChatGPT to get answers. Also, if you already got 900 users, I'd say that's tricky if you consider yourself still a newbie regarding systems administration...

    Thanked by 1totally_not_banned
  • For CPU steal, I would measure it over time before deciding whether it is just a noisy period or a node problem. Inside the VM, check %st with tools like mpstat/sar/top during both idle time and a small controlled load test. Occasional spikes can happen on budget VDS plans, but sustained steal while the VM is otherwise idle is worth a ticket.

    It also helps to run the same benchmark or workload at two different times of day and attach those numbers to the ticket. Since they fixed it for you, I’d ask what changed on their side so you know whether it was migration, node load, throttling, or something else.

  • @forest said:
    Did you open a ticket before opening this thread?

    yes several

  • @woinokiz said:
    I think it's always better to contact host atleast once before coming into forums, as this will stay in records forever

    did that, waited for 6 hours no reply.

  • @dennisadmod said:
    For CPU steal, I would measure it over time before deciding whether it is just a noisy period or a node problem. Inside the VM, check %st with tools like mpstat/sar/top during both idle time and a small controlled load test. Occasional spikes can happen on budget VDS plans, but sustained steal while the VM is otherwise idle is worth a ticket.

    It also helps to run the same benchmark or workload at two different times of day and attach those numbers to the ticket. Since they fixed it for you, I’d ask what changed on their side so you know whether it was migration, node load, throttling, or something else.

    checked it over a two days period, same results. ran few more tests and contacted the support, got no reply so I decided to post here.
    also worth mentioning i couldn't open a ticket to technical department cuz i had a ticket open with sales, when i tried to close the ticket they had a bug in their website that logged me out everytimei tried so had no choice but to post here.

  • sshboxsshbox Member

    @johnwinchester said:

    @woinokiz said:
    I think it's always better to contact host atleast once before coming into forums, as this will stay in records forever

    did that, waited for 6 hours no reply.

    Unless you are paying for 24/7 support, 6 hours isn't much when it's outside of business hours.

    Thanked by 2equalz JohnnySac
  • Support response Under 1 hour

    https://deluxhost.net/en

  • th is going on here?
    why am i getting judged for asking a question?

    Thanked by 1stable_genius
  • zedzed Member

    @johnwinchester said:
    th is going on here?
    why am i getting judged for asking a question?

    situation's resolved anyway, you don't have to respond to all the stupid shit.

  • NeoonNeoon Community Contributor, Veteran

    VDS are not dedicated to you, as we learned from netcup.
    They just put less VPS's on a single node and if the cpu load increases, they live migrate some to other nodes.

  • forestforest Member

    @johnwinchester said:

    @woinokiz said:
    I think it's always better to contact host atleast once before coming into forums, as this will stay in records forever

    did that, waited for 6 hours no reply.

    Bruh, it's a low-end VPS. 6 hours is nothing. Come back to us if it reaches 6 days.

  • lolol @ "I waited 6 hours then came to LET"

Sign In or Register to comment.