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
I just meant if you leave it running at home and it doesn't crash, but it crashes in the DC then maybe it's an environment issue (difference in heat/cold/etc) - but you'd potentially have to leave it running at home for a few days to check this.
@FlorinMarian You should do one of the two:
1. Fix an issue with customers (move them to not-so-yours hardware in the data center); then calmly work to find a root of issue with your hardware.
2. Calmly work to find a root of issue with your hardware. But i doubt any single customer can wait so long.
Regarding the root cause, I'd suggest doing 10 days minimal setup test first. Disconnect all components and do minimal setup with one memory stick, fresh hard drive with W10 installed, re-inserted CPU, i'd recommend to buy or rent another power supply as it is usual root of system unexpected shut down. Run Windows 10 with AIDA64 benchmark with full load on all components on the running server. Be sure to run your hardware with enterprise-grade or professional voltage stabilizer during these tests.
If it still shut down in 10 days of non-stop work with minimal setup, then replace memory stick with another (foreign / new) memory stick and repeat test for 10 days.
If it still shut down in 10 days of non-stop work, then replace components one by one, starting with PSU, then CPU, then MB.
If during components replace you see that you can not replace some components due to missing or not-possible-to-find parts, then i'd suggest to just give hardware to local hardware repair service, as without spare parts it can be impossible to find a root cause.
When it do not shut down, then add components one by one, repeat 10 days test, and see when it will shut down.
did you load must up to date BIOS version?
load BIOS default settings and then review all settings
do you have any repeatability in observed MemTest86 errors?
if so than you could run MemTest in selective mode (go to MemTest configuration and choose only this specific test which fails first, and even you could narrow test address range) ONLY to accelerate speed of finding the root of problem (affected module)
if not sure than start recording these errors details (test number and address)
if you have any other system(s) compliant with these memory modules than use it(them) as well to pre-test these modules (maybe only some of them are broken). it will speed things up, if you could run these tests in parallel
check if all these modules are identical (model name/PN/spare PN...). try to match identical ones
make visual inspection if they are free of some mechanical defects
check/reapply new thermal paste on CPUs
hope you are aware of precautions for dealing with ESD sensitive semiconductor devices…
Thank you for feedback!
Unfortunatelly I have no idea about BIOS because I didn't found any information related to it on HP Firmware database.
About RAM, all performed tests with memtest86 reported 0 errors.
Best regards, Florin.
I've sat down and re-evaluated the situation. The conclusion is: reliability is king and I'm not willing to risk using that server again. It will serve as an 'organ donor'.
I thank everyone who provided advice and helped.
This promo will continue on another hardware rented from the same DC.
Not later than tomorrow we'll start to migrate all the services (0 downtime, proxmox live-migration in action).
Best regards, Florin.
We're happy to announce that we'll have this year black friday offers, being part of the huge list of providers involved in this sale event on LET.
Also we added 2nd server to securely receive load if 1st will cause more troubles.
If before our physical server crashed in 12-26 hours, now it has about 40 hours without downtime.
We found a critical BIOS update which wasn't applied (even it was released by HP in 2019).
Thank you everyone for your cooperation!
This is very common. There's lots of stats that show failures among drives from the same batches failing close together. Real defects show up in first few years but normal drives have roughly the same lifespan, so it's common to see 4+ years old drives from same batch fail within weeks of each other or during a rebuild after the first went tits up. So it's common advice to use drives from different MFG dates to avoid this.
Today only
Use coupon code 80H61FEFUO and get 50% OFF LIFETIME for any VPS SSD/Storage/Game package.
Best regards, Florin.
Hello, dear ones!
We're glad to announce that our VPS packages will be extended due of successful sales from past days .
Without other introductions, we ordered 8x 4TB Drives and another 8x 16GB DDR4 DIMMs.
They will be delivered in about one week and our physical server will have 16/16 DIMMs busy and also 12/12 LFF drives, 2/4 SFF drives. (New Samsung SSDs will be ordered soon).
Thank you for your trust!
Best regards, Florin.
Best of luck to you sir. Personally i run Dell 730xd`s with 2 x xeon e5 2697v3, 384gb memory and 24 x 1tb ssd disks. I got this dirt cheap. Wondering what systems you run actually costs today, as i have pallets of such servers laying in my storage unit
A while ago I got access to two VPSs from the promo, one with 10 GB SSD and one with 50 GB HDD. Both cost about €24 per year. If I'm not mistaken that's the only difference between the two VMs.
As one of the nodes failed repeatedly during the tests I deleted all "old" results and what you get to see here is based on about 50 runs on one and about 150 runs on the other system. Btw, I'm happy to report that since @FlorinMarian installed the new node everything has been running flawlessly and without interruptions.
Thanks, FlorinMarian, for providing the two test VMs.
Here are the results of the SSD VPS, using the second newest vpsbench version (more below):
When looking at those results keep in mind that we talk about VMs that cost less than €25 per year yet come with Voxility anti-DDOS up to 1 Tb/s so evidently it makes little sense to compare those systems to Epyc or Xeon 26xx v4 VMs with NVMe.
CPU and memory performance are decent and a bit higher than what I saw on the many 26xx v2 that are quite common on LET. I'd like to guide your attention to the rather decent (low) spread although the node occupation increased during the time period of the tests. Considering that quite many not at all small dynamic sites run fine and not slow on the somewhat weaker v2 processors I'd call this machine a decent bread and butter work horse albeit due to its modest amount of RAM I wouldn't try to put massive WordPress sites on it. But for (largely) static sites and many other purposes this machine is a decent basis.
Now to some new features of vpsbench albeit small ones: You now get to see whether a system offers hardware AES support, nested virtualization, and hardware random generator support.
On to the drive. vpsbench now not only does the 3 common load sizes, 4KB, 64KB, and 1MB but it also does a (4KB based) IOPS test using 4 threads. One could obviously discuss whether 4 threads do justice to say an 8 vCore system but keep in mind what vpsbench is about! It's much about comparability and not about benchmarking and exploring the depths of each area and for that standard tests are needed. Testing e.g. an 8 vCore systems drives using 32 threads certainly would reflect its disk speed better - but we'd loose comparability.
The results are quite decent for a SSD based system and I like the low spread which boils down to stable and reliable performance. (Hint: but wait for the 2nd test VM ...)
Finally to the network.
Only few failures, and largely where they are expected, so far so good. But only one target, Romania, with close to 100 Mb/s. Yes, the VMs are known to be limited to 100 MB/s as is not uncommon and IMO OK for cheap systems, but all in all those results are nothing to write home about; NL AMS, DE FRA and UK LON in the low to mid 80s and US NYC about 50 Mb/s is "OK" but I'd like to see higher numbers. I mean we're talking about a €2/mo system with 512 MB memory so I'd expect somewhat better connectivity. The 5 TB traffic included in the package are nice though.
Regarding the other system, the one with 50 GB HDD, I'll only present the drive results because otherwise the systems are virtually identical.
Uhm, I don't see a major difference compared to the SSD results. Yes, some numbers are somewhat lower for the HDD system, no surprise there, but IMO those HDD results are quite close and actually great for a HDD based VM.
So I have two pieces of advice/conclusion, one for LET users (and potential buyers) and one for Florin Marian:
@LET users - forget about the SSD box, get the HDD offer. You'll get only slightly less performance but 5 times the capacity!
@FlorinMarian - I'm not sure you did yourself a favour there. Let me word it positively: your HDD setup/config is great, really; "unfortunately" it's almost as good as your (not at all bad) SSD so why should we ignore 80% of the capacity for a not really significant performance gain?
RIP https://proof.hazi.ro/
Do people actually need cheap servers in Romania?
no, we need cheap servers anywhere
I think so, yes, and some success stories of romanian providers confirm that, even cociu's success (although he went up into a clusterf_ck, he still had thousands of customers).
The reason probably is in different factors, ranging from low el. power costs to a liberal attitude (up to and including the - highly likely not properly founded in reality - reputation of not caring much about copyrights and such ("no DCMA")).
Finally, while many European tend to consider Romania some kind of second rate country at the periphery, from a global perspective it makes little difference whether a server is in say Finland, Romania, or the Netherlands.
i need genkbench5 test
No problem, just run it.
https://talk.lowendspirit.com/discussion/comment/76453/#Comment_76453
Incoming UDP is completed blocked due to DDoS mitigation.
It's somehow my fault.
I ran iperf3 tests, which triggered Voxility AntiDDoS to block the port as soon as I increased the rate to 100Mbps.
SpartanHost has a similar block, but it wouldn't kick in so quickly, and I was given the option to choose an unprotected IP when I complained.
Florin asked about unprotected IP but data center won't provide one.
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2021-12-03
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Fri Dec 10 19:20:02 UTC 2021
Basic System Information:
Processor : Intel(R) Xeon(R) CPU E5-2637 v3 @ 3.50GHz
CPU cores : 1 @ 1199.952 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 1.0 GiB
Swap : 0.0 KiB
Disk :
bash: line 422: bc: command not found
fio Disk Speed Tests (Mixed R/W 50/50):
iperf3 Network Speed Tests (IPv4):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 717 Mbits/sec | 468 Mbits/sec
Online.net | Paris, FR (10G) | 733 Mbits/sec | 584 Mbits/sec
WorldStream | The Netherlands (10G) | busy | busy
WebHorizon | Singapore (1G) | 174 Mbits/sec | busy
Clouvider | NYC, NY, US (10G) | 517 Mbits/sec | 559 Mbits/sec
Velocity Online | Tallahassee, FL, US (10G) | 17.3 Mbits/sec | 111 Mbits/sec
Clouvider | Los Angeles, CA, US (10G) | 439 Mbits/sec | 418 Mbits/sec
Iveloz Telecom | Sao Paulo, BR (2G) | 296 Mbits/sec | 516 Mbits/sec
iperf3 Network Speed Tests (IPv6):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 456 Mbits/sec | 445 Mbits/sec
Online.net | Paris, FR (10G) | 449 Mbits/sec | 66.9 Mbits/sec
WorldStream | The Netherlands (10G) | busy | 424 Mbits/sec
WebHorizon | Singapore (1G) | 119 Mbits/sec | 362 Mbits/sec
Clouvider | NYC, NY, US (10G) | 302 Mbits/sec | 384 Mbits/sec
Clouvider | Los Angeles, CA, US (10G) | 350 Mbits/sec | 321 Mbits/sec
Geekbench 5 Benchmark Test:
Test | Value
|
Single Core | 83
Multi Core | 92
Full Test | https://browser.geekbench.com/v5/cpu/11533760
[LXC]VPS Storage - Black Friday