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
Hetzner CX11-CEPH - Nuremberg
Hetzner CX11 - Nuremberg
@benj0x saw the same on an old jessie box... sadly did not have the time yet to check, what might be the cause, libaio maybe...
It was a dedicated (Vultr hourly baremetal). I suspect as @Falzo mentioned that it was just idling at the time the stats were grepped.
I definitely agree on implementing Geekbench 5 to replace 4. This is the next top priority. I will likely add in a switch for anyone who still wants to still use the v4 test instead of the v5 test, but the new default will be v5 once implemented.
Also, that's an interesting idea adding all the geekbench results to a yabs account. I'll have to think about that. I do like just writing out the claim URL to a file so that the user can claim the test and add it to their profile, if desired. I'll think about this further.
Thanks for the bug report! Can you let me know which provider this VM is with so I can investigate further. I fixed the script so that if fio fails, it'll just say it failed rather than printing an empty table. While testing, I did find that Ubuntu 19.10 and Debian 10 templates from Vultr would not run fio (even if I compiled it in the VM itself). Not sure what they have going on there, but I suspect a similar issue in this case.
If the fio test fails, I plan on having it fall back to just doing a simple dd sequential speed test instead, that way at least some disk speed results are shown even if they're not real-world results.
Typically your OS handles which block size to use for I/O operations, which tries to optimize I/O operations given the machine's the filesystem, cache, hardware, etc. With the fio benchmark, an explicit blocksize is given because the "direct" flag is being utilized in order to see the non-buffered disk speed results. As the block size is increased, it can be ascertained from the results if disk throughput or IOPS are the bottleneck for your specific machine. Some additional info can be found in this ServerFault answer.
I'm still a fio noob, so someone with a wealth of more experience with it, such as @Falzo, could probably provide additional insight. The main thing that makes the fio disk tests more akin to "real-world results" are that the I/O operations are random, rather than sequential (which rarely, if ever, occurs in normal server operations).
Thank you for your fast reply.
I'm hosting at netcup and using their default Ubuntu 18.04 LTS Image without any control panels installed.
Thanks for the info. I'll see if there's anything that can be done to make it work. If you have the time, can you download, compile, and run fio manually to ensure it's not something in the VM config or hypervisor that is causing the issue? As long as the OS is glibc-based, the compiled binaries should run, but as we see here that's not always the case.
Until this issue can be resolved, I'll try to quickly implement the dd fallback so some disk speed tests are shown even if fio fails to run.
You're not paying electricity, crank that mother up!
@MasonR nice update sir, works great now
Purchased yabs.sh yesterday. You can now run the test with the shortened URL below --
curl -sL yabs.sh | bash
Redirection is based on user agent. Curl/wget gets redirected to the raw script content hosted on github, while any other agent (i.e. a web browser) gets redirected to the github project page. Enjoy
I noticed that today, very prem indeed sir.
how about adding a simple check if fio is already installed on the system, and prefer a preinstalled version, so you would not even need to bother downloading it? ;-)
I usually have it installed anyway, but even if not and the precompiled would fail the user still could easily decide with a simple apt/yum to make it work again...
btw: small awesome box in toronto from lunanode @perennate
here's what you gave up on @TimboJones - thx again, gonna be a good long term idler in my collection ;-)
Hello guys, I got this error on a new server. Does anyone know how to fix this?
CPU is Intel(R) Xeon(R) E-2174G CPU @ 3.80GHz
curl -s https://raw.githubusercontent.com/masonr/yet-another-bench-script/master/yabs.sh | bash
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-02-10
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Wed Feb 12 22:47:33 UTC 2020
bash: line 29: /bin/uname: Permission denied
Architecture not supported by YABS.
Interesting. Which OS are you using? Output of
ls -l /bin/uname
?Ubuntu 16.04.6 LTS
Output of
ls -l /bin/uname
?It's most likely just a permission problem. Run as your root user for a hopefully quick solution.
This script is awesome!
Nexusbytes (prem) budger series . Bench by Bob.
ZFS Raid 1 with 2xDatacenter NVMe on Hetzner AX61
IOPS seems weird
I have a KVM VPS with Ubuntu 20.
I tried to run this script with:
curl -sL yabs.sh | bash
but got this message"
-bash: curl: command not found
What should I do now?
apt install curl -y
Then, run the benchmark script.
Don't forget to run update & upgrade before the install.
Sometimes you may get funny responses if you used an (older) 20.04 template
cancel your unmanaged vserver and buy shared hosting.
On a good day, I would have answered something similar.
Now this made me laugh
This tool looks great.
Testing a Black Friday KVM offer from @VPSSLIM
How long should the
Running GB5 benchmark test... *cue elevator music*
ticker stay? It's been up for 30 minutes already.If it decides to continue (so far
fio Disk Speed Tests
are less than promising) I'll post my results.Well, you should take a look at the other @VPSSLIM test and see that the geekbench score is a complete garbage. It performs even worse than the CPUs offered in HostSolutions storage servers, which are already the pain in the ass.
If you really want to host something productive on them: don't do it. The performance is terrible.
I was testing them, since they put up an appealing offer for 3 EUR, but it doesn't even run a Minecraft server.
Welp, as they say, you get what you pay for... Lesson learned.
@VPSSLIM benchmark test for KVM #2:
2x Intel Xeon CPU
4 GB RAM
150 GB SSD
5 TB bandwidth
Retail price at 17.95 EUR / mo, offered in Black Friday deal for 3.05 EUR /mo, tax included
I also had tons of problems setting it up, making it work, it's been 2 weeks and still can't run anything in it because of the "High I/O" alert suspends it.
Lesson: learned.
Don't think I've ever seen a Geekbench as low as that.