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
Please see https://wiki.hetzner.de/index.php/CloudServer/en#When_will_I_get_my_invoice.3F for this and other frequently asked questions. --Katie, Marketing
There's not much that I can say here because I don't have access to those systems. Sometimes an initial order is approved of by our system, but then when our technicians have a chance to review the orders more closesly a day or two later, they may find suspicious details that the software missed. I am guessing that this is what happened in your case.
I cannot do anything if your order has been rejected. There are a number of reasons why this may have happened. For example, perhaps you gave us incorrect information when your ordered your server, or perhaps you do not have a valid ID. You can try to make a new order, and make sure that your information is completely correct.
We reject orders not to make you feel bad, but to protect our customers and everyone else on the Internet. Please understand that we do not mean to personally reject anyone. If an order looks suspicious, we will reject it to protect everyone else, even if that means losing business. --Katie, Marketing
Just switched with a big site onto Hetzner (2-Core-40-GB-NVMe-Local). 15 Million hits monthly. Cloudflare is in front, hosted on NginX (1.12) + php-fpm7.0. SQL @ Hetzner on another machine (same specs)
Before: OVH dedicated server - Raid 10 (no ssd)
So after all: Double to Triple Bandwidth and half / third of the load time. Looks great so far!
Update: Having ~ 40 % usage on the web and 20 % usage on the SQL VM. Seems solid.
Just curious, where you testing from Falkenstein or Nuremberg? Any speed difference between the two for you? (Using your test file I'm getting ~200MBs to Falkenstein, 24 seconds to download)
Well its faster than my disk (NVME ... I think)
https://serverscope.io/trials/969X
Quite the jump in IOPS from a 2GB to 4GB server.
Are you missing another chart? because the two graphs in the image are measuring different things (first speed, second response time)
Take a look at the end. the bandwidth usage/speed has increased and the response time has dropped down.
@Hetzner_OL I created my first two instances and my first impression of the performance is very good.
I know it's impossible to predict the future. Anyhow - is the current spec-pricing-scheme somehow an 'introductory offer' or is it here to stay?
I'm looking for some stability after I have been busy shifting around instances during the last years...
How easy is it to load a snapshot/image from another provider into their system?
I expect to be able to load an image, use something like System Rescue CD to change the IP addresses and perhaps some grub settings (this shouldn't really be necessary) and get it going.
There is no image import (yet). Maybe later. Just make a backup and go with cloud-init to prepare your machines with that (its like Anaconda).
Further upthread they said it is not introductory. I don't see how the performance can last, but we'll see.
As a follow-up to my previous posts regarding Hetzner->Linode transfer speeds, I actually went ahead and tried the speed from every Linode location. I compared against the 1GB test file from the Vultr Frankfurt location:
wget -4 -O /dev/null http://speed.hetzner.de/1GB.bin wget -4 -O /dev/null http://fra-de-ping.vultr.com/vultr.com.1000MB.binResults:
Fremont: Hetzner: 110KB/s (2h35m) Vultr: 16.4MB/s (1m1s) Dallas: Hetzner: 13.1MB/s (1m16s) Vultr: 16.0MB/s (1m2s) Atlanta: Hetzner: 8.06MB/s (2m4s) Vultr: 23.6MB/s (0m42s) Newark: Hetzner: 21.3MB/s (0m47s) Vultr: 28.1MB/s (0m36s) London: Hetzner: 78.2MB/s (0m13s) Vultr: 147MB/s (0m6.8s) Frankfurt: Hetzner: 110MB/s (0m9.1s) Vultr: 197MB/s (0m5.1s) Singapore: Hetzner: 8.25MB/s (2m1s) Vultr: 13.3MB/s (1m15s) Tokyo: Hetzner: 7.31MB/s (2m17s) Vultr: 10.9MB/s (1m32s)This pastebin contains a traceroute for each location: https://pastebin.com/UNv3W54j
Can you also do reverse traceroute for Fremont (the most problematic one)? Thanks.
Added: my measurements:
Hetzner 1GB file to Virmach.com (San Jose, California): 965KB/s, 29m 16s. Speed varied a lot during the transfer, sometimes up to several MB/s, sometimes in the 200kB/s range or maybe even lower.
Vultr: 17.4MB/s, 58 sec. OMG. Maybe I should proxy my Hetzner servers through there. I've always just figured Hetzner's transatlantic routes weren't very good but this is an amazing contrast. I don't know if I should open a ticket with Hetzner, given that they're probably already aware of the issue.
.
I don't have an active Hetzner server and I can't find an official Hetzner Looking Glass. Googling turned up http://lg-de-fks.mrvm.net/ which is hosted at Hetzner's Falkenstein DC. The speed test server seems to be at Nuremberg so it's not ideal, but the best I can find.
The very poor transfer speed yourself seeing sounds similar to what I've experienced, including huge variance during the download. The transfer pricing from Hetzner is fantastic but it's just not worth it if I can't reliably get 10MB/s to the US.
traceroute to speedtest.fremont.linode.com (50.116.14.9), 30 hops max, 60 byte packets 1 virt-de01.mrvm.net (88.198.69.215) 0.052 ms 0.024 ms 0.019 ms 2 core21.fsn1.hetzner.com (213.239.229.121) 0.314 ms 0.294 ms core22.fsn1.hetzner.com (213.239.229.125) 0.301 ms 3 core1.fra.hetzner.com (213.239.245.218) 5.164 ms 5.150 ms 5.132 ms 4 core2.ams.hetzner.com (213.239.203.158) 10.603 ms 10.600 ms 10.637 ms 5 30gigabitethernet1-3.core1.ams1.he.net (80.249.209.150) 10.753 ms 10.770 ms 10.663 ms 6 * 100ge9-2.core1.par2.he.net (184.105.81.109) 20.803 ms 19.996 ms 7 100ge7-1.core1.lon2.he.net (184.105.223.253) 30.724 ms 32.289 ms 34.399 ms 8 100ge4-1.core1.nyc4.he.net (72.52.92.166) 91.280 ms 88.862 ms 90.300 ms 9 100ge8-1.core1.sjc2.he.net (184.105.81.213) 153.114 ms 150.406 ms 150.149 ms 10 100ge4-2.core3.fmt2.he.net (72.52.92.245) 153.529 ms 165.913 ms 154.038 ms 11 * * * 12 173.230.159.17 (173.230.159.17) 152.584 ms 173.230.159.13 (173.230.159.13) 152.921 ms 173.230.159.17 (173.230.159.17) 152.571 ms 13 speedtest.fremont.linode.com (50.116.14.9) 152.356 ms 154.081 ms 150.629 msThats bad configured routing on hetzners side. they have every good upstream to .us like tata, level3, gtt, ntt and telia but the routes they have configured are crap.
they just need to fix the routing
Is it possible to install Ubuntu 12.04 on Hetzner dedicated servers?
12.04 will be EOL at the end of April.
Is it any difference if I order dedicated server from Hetzner right now or on 1st February? Should I pay for the rest of January?
You'll get charged the pro-rated amount for January. It all works out. If you want the server now, buy it now.
If they do go ahead and improve routing that would be great. My concern is two-fold:
1) The fact this issue exists and has been a problem form the past few days I've been trying to evaluate Hetzner using their speed test isn't great
2) There's no indication this is a known issue that's been caught by their own tracking. Would I need to set up my own externally hosted infrastructure to double check transatlantic speeds aren't an order of magnitude or more below what you'd expect?
Just spin up the new server before removing the old one
Defeats the purpose of hourly servers which is to spin them down when you're not using them. Admittedly this was more of an issue when I was spinning a lot of servers up and down during testing. I normally wouldn't do that so much.
Finally i've done the successfull installation about Windows on Hetzner cloud. Works well and very very responsive. i love Nvme
Please let us know how you did it
Hm... interesting.
I was allowed to create a server without paying anything in advance. ( a few days ago )
I did receive an invoice today, though. I paid through paypal and no ID was requested.
falkenstein vs nurnberg ?
Ordered NVME dedicated server, got it in falkenstein
I think nurnberg is better location, do you think so? Is it a way to change dc
Define ' better location ' o_o
I've tested a bit, overall, pings are lower to nurnberg. Also, I think nurnberg is a bit city, falkenstein is a small "village". Anyway, maybe someone had real experience and could compare.
I'm still seeing really poor transfer speeds from Hetzner to Linode Fremont. In the end, I actually launched a Hetzner VM and compared it to transfer from my @Clouvider VM, and a DigitalOcean VM in Frankfurt. Yesterday I was seeing slow speeds from Clouvider/DO Frankfurt/Hetzner Nuremberg to Linode Fremont. Today Clouvider is speedy, but DO and Hetzner remain slow. Traceroute output is unchanged (@Clouvider - have you changed anything in the last day or so?).
iperf transferring data from Hetzner Nuremburg to Linode Fremont [ 3] local 195.201.17.2 port 52910 connected with 45.33.39.8 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 2.12 MBytes 1.78 Mbits/sec [ 3] 10.0-20.0 sec 640 KBytes 524 Kbits/sec [ 3] 20.0-30.0 sec 640 KBytes 524 Kbits/sec [ 3] 30.0-40.0 sec 256 KBytes 210 Kbits/sec [ 3] 40.0-50.0 sec 640 KBytes 524 Kbits/sec [ 3] 50.0-60.0 sec 640 KBytes 524 Kbits/sec [ 3] 0.0-66.6 sec 5.00 MBytes 629 Kbits/secClouvider results are 145MBit/sec this evening, but earlier today weren't great for this particular route:
iperf transferring data from Clouvider London to Linode Fremont [ 3] local 185.42.223.54 port 56214 connected with 45.33.39.8 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 6.00 MBytes 5.03 Mbits/sec [ 3] 10.0-20.0 sec 1.88 MBytes 1.57 Mbits/sec [ 3] 20.0-30.0 sec 2.00 MBytes 1.68 Mbits/sec [ 3] 30.0-40.0 sec 2.50 MBytes 2.10 Mbits/sec [ 3] 40.0-50.0 sec 2.00 MBytes 1.68 Mbits/sec [ 3] 50.0-60.0 sec 2.50 MBytes 2.10 Mbits/sec [ 3] 0.0-60.8 sec 16.9 MBytes 2.33 Mbits/seciperf transferring data from DO Frankfurt to Linode Fremont [ 3] local 159.89.7.132 port 33754 connected with 45.33.39.8 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec [ 3] 10.0-20.0 sec 384 KBytes 315 Kbits/sec [ 3] 20.0-30.0 sec 1.00 MBytes 839 Kbits/sec [ 3] 30.0-40.0 sec 640 KBytes 524 Kbits/sec [ 3] 40.0-50.0 sec 512 KBytes 419 Kbits/sec [ 3] 50.0-60.0 sec 896 KBytes 734 Kbits/sec [ 3] 0.0-60.6 sec 4.62 MBytes 640 Kbits/secComplete results including traceroute are here, including results for EU->DO SFO for good measure: https://pastebin.com/B5MLMx8s
No changes our end. We handover to their upstream at LONAP, our LONAP port peak consumption over past 7 days was 11% so it’s either congestion on the other end or their transit that has congestion somewhere