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
Searching IP assigned to this VPS on Google shows "My Server Planet LLC" and ASN "AS36352 ColoCrossing"
Lol ohh god, someone update that immediately
Well, whatever location you're at has IPv6 connectivity as it passes the connectivity check done as part of the script:
@MasonR solved my issue. I am running CSF and had to add ports 5200-5210 to my TCP_Out rules.
Awesome, thanks @MasonR
Here's my Online.net 2.99/month server:
I think I win the read/write speeds so far (no it's not NVME nor cached).
I'd like to see some real world simulated tests that helps compare servers, like max concurrent nginx users, or max wireguard users, etc.
Also, the world really needs a better replacement for iperf. The more one uses it and sees all the bugs, the more you want to throw it against the wall and stomp on it.
Can we start archiving the results someplace other than on LET threads and geekbench.com?
That's something that @eva2000 excels at with his centminmod benchmark. Pretty sure he has a script you can run to do all that good stuff and see some real-world results.
Awesome stuff @MasonR
For years to come families can now play "Guess what I'm idling" every Christmas.
My magical BuyVM slice in Las Vegas just barely beat out your read speed. Thanks @Francisco!
send nudes
What kind of SmallOS doesn't have man installed? :P
FeministVM.
VirMach LA 4G Special
But yet
IPv6 check seems to be broken?
Nice can never have enough benchmark/testing ^_^ Though DD and sequential disk I/O isn't that indicative compared to random read/writes i.e. fio and sysbench. I wrote a generic sysbench benchmark script at https://github.com/centminmod/centminmod-sysbench you can draw some inspiration for it you want
Yeah my centminmodbench.sh script is more tailored to real world for some tests but it's specifically written for centos and centmin mod lemp stack environment heh
Anyway I had opportunity to test run YABs on a fresh dedicated server setup as my old databasebydesign E3-1270v1's 240GB SSD disk died and they replaced the server with E3-1230v2 240GB SSD. Loaded up CentOS 7.7 64bit with Centmin Mod latest beta LEMP stack and then ran your YABs script.
Prem benchmark @MasonR I shall be using this from now on. Great Work!
There are 2 problems with this script with me,
1- Server have 2x500gb HDD 7200 rpm sata3 nothing special but in ZFS RAIDZ1
This dd speed test result is not true...
and also same as @sanvit server have no ipv6 but I can see ipv6 test results.
Yeah, something is off. I'll get it debugged, tested, and patched later today.
Thanks for testing the script out! I agree dd isn't a truly accurate measure. Seems to be the biggest complaint so far, but I haven't found a suitable alternative which won't require external dependencies to be installed/elevated privileges. If you know of one, do tell! Goal is to keep the script as lean and dependency-free as possible -- once something needs to be installed prior or built/compiled to run the test, I've failed my goal :P
Yep, this has been addressed several times over now. What I'll probably end up doing is mimic nench and use ioping for read tests and dd for write tests. Should give a more accurate measure than dd's read.
I'll get the IPv6 stuff figured and patched up at some point today.
Thanks @MasonR . Quality contribution to the community. As usual.
Posting snippets of results from Hosthatch Chicago (after the node replacement / hardware changes over the weekend) and Hostdoc, Finland (HDD). I've put together results from 4 YABS tests here Other two results include Ramnode Dallas - Tinykvm and Stockservers Germany SSD)
So far, the Stockservers VPS took nearly 20 minutes to run the BM,
surprisingly the slowest of the lot.
**Hosthatch Chicago **
Mon Oct 7 16:08:57 IST 2019
Basic System Information:
———————————
Processor : Intel(R) Xeon(R) CPU E5-2690 v2 @ 3.00GHz
CPU cores : 1 @ 2999.998 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ❌ Disabled
RAM : 987Mi
Swap : 1.0Gi
Disk : 15G
dd Disk Speed Tests:
———————————
| Test 1 | Test 2 | Test 3 | Avg
| | | |
Write | 185 MB/s | 235 MB/s | 243 MB/s | 221 MB/s
Read* | 311 MB/s | 308 MB/s | 319 MB/s | 312.667 MB/s
HostDoc - Finland (HDD)
Tue Oct 8 20:47:04 IST 2019
Basic System Information:
Processor : Intel(R) Xeon(R) CPU E3-1246 v3 @ 3.50GHz
CPU cores : 3 @ 3499.996 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ❌ Disabled
RAM : 1.9Gi
Swap : 2.0Gi
Disk : 28G
dd Disk Speed Tests:
Write | 105 MB/s | 82.7 MB/s | 118 MB/s | 101.9 MB/s
Read* | 89.7 MB/s | 127 MB/s | 61.6 MB/s | 92.7667 MB/s
iperf3 Network Speed Tests (IPv4):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Bouygues Telecom | Paris, FR (10G) | 818 Mbits/sec | 922 Mbits/sec
Online.net | Paris, FR (10G) | 817 Mbits/sec | 899 Mbits/sec
Severius | The Netherlands (10G) | 822 Mbits/sec | 767 Mbits/sec
Worldstream | The Netherlands (10G) | busy | busy
wilhelm.tel | Hamburg, DE (10G) | 824 Mbits/sec | 778 Mbits/sec
Biznet | Bogor, Indonesia (1G) | 0.00 bits/sec | 0.00 bits/sec
Hostkey | Moscow, RU (1G) | 832 Mbits/sec | 937 Mbits/sec
Velocity Online | Tallahassee, FL, US (10G) | 741 Mbits/sec | 399 Mbits/sec
Airstream Communications | Eau Claire, WI, US (10G) | 703 Mbits/sec | 91.1 Mbits/sec
Hurricane Electric | Fremont, CA, US (10G) | 736 Mbits/sec | busy
iperf3 Network Speed Tests (IPv6):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Bouygues Telecom | Paris, FR (10G) | 814 Mbits/sec | 923 Mbits/sec
Online.net | Paris, FR (10G) | busy | busy
Severius | The Netherlands (10G) | 823 Mbits/sec | 840 Mbits/sec
Worldstream | The Netherlands (10G) | 819 Mbits/sec | 926 Mbits/sec
wilhelm.tel | Hamburg, DE (10G) | 823 Mbits/sec | 928 Mbits/sec
Airstream Communications | Eau Claire, WI, US (10G) | 704 Mbits/sec | 112 Mbits/sec
Hurricane Electric | Fremont, CA, US (10G) | 722 Mbits/sec | busy
Awesome, great write-up! Thanks for testing!
A lot of things can affect the time it takes to run the script. The biggest factor I've seen is if the iperf servers are getting hammered when you run test and it'll have to do a bunch of retries per location to finally connect. I'll probably eventually put in a flag to let you specify the number of retries (default is 10) so the script doesn't take as long, if that's something people want.
great contribution @MasonR - as always!!
+1 for fio instead of dd. would like to see 64k bs and 4k IOps and readwrite speed from that ;-)
probably also need to look out for more iperf locations to add to the list ^^
tl;dr - now using ioping for disk reads (less tainted by cache), fixed IPv4/6 detection, todo list seems to be growing as fast as it's shrinking.
Alright, so I just pushed a big update with a slew of changes, including, but not limited to:
@MissFortune - if you get a chance, can you retest this on your CentOS machine? You should no longer get the "permission denied" error as long as you have write permissions to the directory that you run the script from
@alilet, @sanvit, @emre - if you guys get a chance, can you retest this to make sure the IPv4/IPv6 checks are working correctly? IPv6 tests should no longer run if your machine doesn't support it.
There's always more to do, but if anyone has any further suggestions and/or bug reports, please keep them coming! Here's my current to-do list (not in any particular order):
This updated fixed the issue with no data about IO
ipv6 Check now working as expected.
but disk speed stuff goes off the roof
Ruh roh raggy. That's unexpected, will have to look into that.
Edit: Seems to be your locale settings throwing it off, I guess. You Europeans using decimal commas instead of decimal points... tisk tisk tisk. Doesn't explain the ludicrous read speeds, though. I guess ioping is more prone to caching than that I had originally thought.
This is a local SSD. Intel D3-S4510 to be exact.
CrownCloud - 4 GB RAM - 60 GB SSD - KVM at Los Angeles
Latest test:
Hi,
I think speedtest.net is a better value for real visitors than iperf
MasonR has done great work. thank you for the code!