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
Yea, got Marseille refunded already.
Here is YABS for the OVH VPS-1 in NY.
It is $6 a month which is pretty nice for 4 EPYC Genoa cores and 8GB RAM.
From what I can tell, it appears this is in the Equinix NY6 data center in Secaucus, New Jersey.
What I don't understand is the disk speed. I guess IOPS are limited? It is advertised as NVMe but you get HDD speeds so that severely handicaps this VPS.
The port is 400Mbps which is not the end of the world but why not just make it 1G? I'd prefer a bandwidth cap that is burstable up to 1G or 10G.
The IPv6 is a /128 which is really lame. Minimum should be a /64 with /48 preferred for dedicated or VPS servers.
The CPU performance is awesome but my main concern is the disk. If the disk speed is going to stay like this then I don't see a point in keeping this VPS.
nws.sh global test:
Are local zones based on Hyper-V?
Correction from OGF: Local Zone VPS has anti-ddos
We're soooooo back
Don't think so. They have probably just enabled Hyper-V enlightenments which makes Windows guests run more efficiently but it is still KVM.
anyone share subnet of madrid pls, also it safe to try their services and get refunded if i dont like it?
Thank you
Edit: https://smokeping.ovh.net/smokeping i found this
VPS 1 in The Netherlands is literally twice as fast with encoding (h264) as the VPS 2 in Germany.
Geekbench single core score for the VPS 1 in The Netherlands is also ~2100 where the VPS 2 in Germany has a ~1000 single core score.
deleted
@Neoon can you tell me the last hops between ovh-vie (e.g. vie-vie2-sbb1-8k.at.eu) and your vps in vie?
Sth. is crazy here with the last hop before the vps not by the internal ovh network but an Arelion AS1299 one* (after reaching the ovh-vie net) and the vps not reachable for ssh but pingable? Never seen that crazy kind of behavior with an ovh vps.
*) Edit: ah, it's not by Arelion but an OVH address named arelion.as1299.es.eu (that sounds like a hop to connect ovh to arelion in Spain, IP seems to be located in Brussels (might be wrong or anycast) but makes all no sense for Vienna. Esp. not with ping times of 300ms after that.
Edit2: Nevermind, they seem to have those 57.128.x.x ips on all localzone routes, just a miss-named IP. So this is probably unrelated. Will see what OVH finds out about Vienna, maybe just a borged OS template there...
Edit3: so I finally found the virtual KVM for VPS at OVH (which is located in a drop down menu where you would not expect to find it) - result: KVM cannot connect, sth. went wrong. Seems to be a really borged host node I got.
Someone of you testing AntiDDOS?
"Anti-DDoS protection enabled for IP address 141.227.xxx.xxx"
Yes but IO are too capped on local zone VPS. No need for insane speeds,
I always feel some people complain too much since it works for most use cases.
But here it's still a bit limited
Good for what?
Yes, the disk speed is pretty low compared to the one in Germany, but in my case I don't need those speeds anyway as it's just being used for 24/7 encoding, only heavy on the CPU.
Hm, Vienna is a bad apple compared to the other vps in the other local zones. Something is seriously broken on that one, and it looks even like an underlying hardware problem. Maybe got a host node that is located on Mars and not in central Europe...
Good test case to see how OVH will handle this (while not expecting too much).
iowait at constantly 25% with zero cpu usage.
Edit: nope, that was just a small timeframce of some good seconds.
iowait is actually at 50% constantly with zero cpu (or other) usage.
You're probably hitting the IOPS throttle. iowait is the percentage of time the CPU is idle because it’s waiting on disk I/O to complete. They set the limit so low that basically any disk activity will cause iowait.
But: I'm not writing anything to disk nor reading. This was while just running iostat, nothing else. Load at 20% (sic). So I had no disk activity really besides maybe syslog logs being written. There is no load whatsoever produced by me on that vps.
And these DL speeds are also not normal, esp. with bbr congestion control:
When you see the iowait spike up to 25% or 50% again, run
iotopto see exactly what process is accessing the disk. If the read/write rate is tiny (but still showing large iowait) then maybe something is wrong with the node you are on.Looks like it is not disk related but cpu related. See no disk activity but 25% to 75% iowait on CPU (-c). That host node thinks I should not get any cpu from the sheduler even while I don't use the cpu at all.
Well I was looking for more ram, I´ve got 6 wordpress websites running, plus a few test wps and owncloud , so I got the VP2-2 6 vcpu 12GB of Ram 100GB SSD, for $7,70 (plus 0,80 for snapshots) and it has free 1-day retention backups.
But heres the catch (And I´ve paid for it already, and I need to cancel my renewal on my AMD Epyc 4vcpu 4GB till september 18, causa I dont want to be stuck with a vps provider, despite being quite a good one in this case for another year )
But the crazy thing is despite the VPS-2 being much slower, I´ve gone from a 700ms cache to a 49ms one in wordpress with the Intel VPS. I Just wish they had put the specs on the website.
Basic System Information:
Uptime : 0 days, 11 hours, 22 minutes
Processor : Intel Core Processor (Haswell, no TSX)
CPU cores : 6 @ 2394.454 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 11.4 GiB
Swap : 0.0 KiB
Disk : 96.8 GiB
Distro : Ubuntu 22.04.5 LTS
Kernel : 5.15.0-135-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online
VPS Benchmark Comparison (Sysbench CPU)
VPS 1 (old, first test)
CPU: AMD EPYC Milan, 4 vCPU
RAM: 4 GB
Single-core (1 thread): 3989.40 events/s
Multi-core (4 threads): 14629.47 events/s
Multi-core / Single-core ratio: ≈ 3.67x
VPS 2 (new, second test)
CPU: Intel Core Processor (Haswell, no TSX), 6 vCPU @ 2.39 GHz
RAM: 11.4 GB
Single-core (1 thread): 946.69 events/s
Multi-core (6 threads): 5269.11 events/s
Multi-core / Single-core ratio: ≈ 5.56x
Analysis
VPS 1 (AMD EPYC) → significantly faster per-core performance, making it better for workloads that rely on single-thread speed (databases, web APIs, lightweight apps).
VPS 2 (Intel Haswell) → much weaker per-core, but shows better scaling efficiency across threads (ratio 5.56x vs 3.67x). It can handle more parallel tasks but is slower overall.
Basic System Information:
Uptime : 2 days, 8 hours, 16 minutes
Processor : AMD EPYC-Milan Processor
CPU cores : 4 @ 2645.030 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 3.7 GiB
Swap : 4.0 GiB
Disk : 77.5 GiB
Distro : Ubuntu 22.04.5 LTS
Kernel : 5.15.0-140-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ❌ Offline
IPv4 Network Information:
ISP : OVH SAS
ASN : AS16276 OVH SAS
Host : OVH Hosting, Inc.
Location : Beauharnois, Quebec (QC)
Country : Canada
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda1):
Basic System Information:
Uptime : 0 days, 11 hours, 22 minutes
Processor : Intel Core Processor (Haswell, no TSX)
CPU cores : 6 @ 2394.454 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 11.4 GiB
Swap : 0.0 KiB
Disk : 96.8 GiB
Distro : Ubuntu 22.04.5 LTS
Kernel : 5.15.0-135-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda1):
How do you have that storage speed on the localzone?
That is not a local zone one but the comparison of a new vps-2 with an old vps-le-amd-epyc (see Milan and processor speed) in BHS (OVH DC in Canada), I would say.
The new local zone vps have Epyc 4th gen and heavily throttled disk i/o.
Yeahp, It´s hosted on their DC in Canada. I think despite the old and slower CPU I will keep it, I dont wanna waste the money and I dont wanna commit to another full year just to keep the AMD Epyc server.
My whole setup is running kind good, so no complains here so far. Have ram to spare always helps.
Ahh damn the cancel option was bugged now I see I could just cancel the commitment.
The vps in local zone VIE is still unusable after 6 days. All management tools via OVH panel do not work or need 7 to 12 hours to complete (e.g. boot into rescue).
Packet loss at 50% + with latency at 2 seconds (sic) up to 60 seconds (really sick).
CPU iowait at 75% to 100% with 0.0% load on system and user. Kernel throwing bugs because it does not get any time from the CPU scheduler. (OVH U2404 template that is running fine on all other local zones)
Ah, and ovhcloud support says: anything is fine, we don't see a problem.
edit:
Was fine before, looks like VIE is fucked...
So your one is playing crazy as well? Mine had been a dead joke since the beginning (I think Thursday or Friday last week Edit: no the craziness started on Sunday or Monday, I think, or to be more precise: night from Saturday to Sunday).
Wow, new state at Vienna: not under maintenance anymore, it's in active state again. Now only the disk and the network is dead and it looks like no one at OVH want to get near that host machine or hypervisor ...
I couldn't even SSH into VIE, so yea..
Maybe I should tiket and also ask for a refund on this one.
Open a support ticket if you want to have some fun. Or open a refund ticket if you want to keep your sanity.
Or open the noVNC KVM console if you want to experience some errors you've never seen before. Might be though that the noVNC KVM console also cannot connect anymore, I now know that state as well.