Howdy, Stranger!

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


How do cheap VPS provider handle hardware failures?
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.

How do cheap VPS provider handle hardware failures?

If your dedi rented from OVH/Hetzner fails they replace part in less than hour, they have everything in stock. If they dont have 1TB drive, they'll give you 2TB like we see with Intel Atom dedis in OVH Eco/Kimsufi.

But what if there's CPU or motherboard failure on LET provider dedi which is colocated somewhere where they don't live?

Netcup colocates in Hetzner, but they said they have spare parts in stock in DC, they can replace them in minutes with remote hands. Downtime is minimal and they have capacity to migrate the VPSes if something can't be fixed fast.

Oracle does live migrations on their VPSes, your instance will just run on other physical server, but with same data.

But how do LET companies do it? Everyone has spares or they don't and you need to wait couple of days in order to fix? Is extra capacity for migrations a normal thing or something you'll never see at LET prices?

I didn't have any hardware failure on any of my dedis/VPSes, so I'm curious how its handled if that will happen. I mean, I had one mail from Oracle that there was "CPU failure" but because of their live migrations I didn't even saw any downtime.
I saw whole situation with Virmach where VPSes aren't working properly for months and that made me think if all LET companies can have such "breakdown".

If you're provider feel free to reply how do you handle hardware failures :)
If you had failure on rented/colocated dedi then let me know how it was handled :)

Thanked by 1jugganuts

Comments

  • hostikohostiko Member, Host Rep
    edited October 2022

    As a provider in a country with an ongoing warfare, it is really important to have facility somewhere outside of our country in case some real shit happen.

    We lease servers at Hetzner for our Germany KVM hypervisors, so if something goes wrong in our Kyiv DC (it is collocation), we could migrate servers to Hetzner, with preserving of IP addresses (we use BGP to provide own networking).

    We are too small to have lots of major hardware components (like CPU) in reserve, but having a few of spare servers after client abandon its prolongation is not so expensive thing to do, especially if you are growing.

  • @AXYZE said: Netcup colocates in Hetzner, but they said they have spare parts in stock in DC, they can replace them in minutes with remote hands. Downtime is minimal and they have capacity to migrate the VPSes if something can't be fixed fast.

    I believe netcup does live migrations if they have planned maintenance. There will of course be some downtime if there's unexpected hardware failure no matter the setup. But anyway the live migrations usually rely on some type of setup where they write your data on disks in another server and if your server is down, they will provision it somewhere else.

    @AXYZE said: But how do LET companies do it? Everyone has spares or they don't and you need to wait couple of days in order to fix?

    You will get vastly different responses from different providers and I am not a provider, but it is common for people colocating to have spare raid cards, ram and disks and you can pay for remote hands to deal with the problem. For things like cpu/mobo, you usually wouldn't have spares unless you have entire racks of the exact same hardware (which is feasible for netcup because they are mostly the same hardware)

    @AXYZE said: Is extra capacity for migrations a normal thing or something you'll never see at LET prices?

    This depends on the scale the provider runs at and how many of their servers fail. I have seen many different responses to hardware failure on LET. Some providers rent servers from elsewhere while trying to resolve the problem. Some start ordering replacement parts and customers will have to wait. (I remember there's a relatively new host on LET who had hardware failure in the last year or so and he only had 1 server at that point in time. He ordered the replacement cpu and everyone had to wait, if I recall correctly.)

    Thanked by 3Havy AXYZE hyena56
  • HavyHavy Member
    edited October 2022

    @AXYZE said:
    If your dedi rented from OVH/Hetzner fails they replace part in less than hour, they have everything in stock. If they dont have 1TB drive, they'll give you 2TB like we see with Intel Atom dedis in OVH Eco/Kimsufi.

    But what if there's CPU or motherboard failure on LET provider dedi which is colocated somewhere where they don't live?

    Netcup colocates in Hetzner, but they said they have spare parts in stock in DC, they can replace them in minutes with remote hands. Downtime is minimal and they have capacity to migrate the VPSes if something can't be fixed fast.

    Oracle does live migrations on their VPSes, your instance will just run on other physical server, but with same data.

    But how do LET companies do it? Everyone has spares or they don't and you need to wait couple of days in order to fix? Is extra capacity for migrations a normal thing or something you'll never see at LET prices?

    I didn't have any hardware failure on any of my dedis/VPSes, so I'm curious how its handled if that will happen. I mean, I had one mail from Oracle that there was "CPU failure" but because of their live migrations I didn't even saw any downtime.
    I saw whole situation with Virmach where VPSes aren't working properly for months and that made me think if all LET companies can have such "breakdown".

    If you're provider feel free to reply how do you handle hardware failures :)
    If you had failure on rented/colocated dedi then let me know how it was handled :)

    For these reasons and others that why you need to be familiar with your provider, have good relationships, etc.

    Seriously I even known a provider (not from here) whose their colocation provider/data center seemed to have A and B feeds only on their clients' servers and not the data center's networking equipment (only an A feed and they took well over an hour to get the generators going followed by fixing issues with the routers and such losing said power).

    So while my dedi rental when I had it didn't dropped power, the networking did drop because the data centers' routers and such didn't have power.

    Thanked by 1AXYZE
  • HostEONSHostEONS Member, Patron Provider
    edited October 2022

    We usually keep spare SSD, Cables, PSU, RAM, CPU, CPU FAN ... moreover we have arrangements even with our colo provider to source us the hardware if needed, we may have to pay a little extra, but it works .. Amazon Prime helps a lot with same delivery at least for common stuff like SSD, ... also usually in our more popular locations we usually keep a spare node available in case of emergency

    Thanked by 2AXYZE let_rocks
  • My experience good LET provider move your vps just to another host machine within 1 day, bad don't care at all and your vps will be down weeks sometimes :(

    Thanked by 1AXYZE
  • If its dirt cheap then its simple. They don't :D and disappear.

  • servarica_haniservarica_hani Member, Patron Provider

    for VPS we always run in pools of servers with n+1 or n+2 redundancy
    so if any server fail or if we need to do upgrade of the server we simply live migrate users to other servers and do the work on the server with issue

    thats why all our storage is SAN so when moving the vm we dont have to move the data

    for our hardware failures or disks we have some spares in the datacenter and some in the office and we do the replacement ourselfs (since we run VMs in pools we rarly has a case that we must fix the server immediately and usually fixing the server can wait 24 hours or more since it does not affect any VMs)

    Same thing for disks as we run raidz2 if a disk fail we replace it ourself without need of remote hands since it is not urgent

  • caiicaii Member
    edited October 2022

    Rebranding and black friday

    Thanked by 2ralf AXYZE
  • As long as it's not a resale VPS, the local operator will fix it quickly.

    Thanked by 1AXYZE
  • Cheap VPS Providers do regular server monitoring and also update their servers regularly. this help a lot to prevent hardware failure.

    Thanked by 1AXYZE
  • bdlbdl Member

    they open tiket and refound

  • They migrate 10 Epyc nodes onto one 4-core Intel Atom based machine because everyone's idling them anyway and nobody will notice.

    Thanked by 2fadedmaple crilla
Sign In or Register to comment.