Speedtest.net for unRAID 6.1+


dmacias

Recommended Posts

I have better results with speedtest-cli:

 

                    Ping -    DL    -    UL

Browser:      3ms - 213.13 - 210.75

Plugin:          7ms - 208.58 -  50.99

stest-cli:      7ms - 208.30 - 203.80

stest-linux:  12ms - 205.12 - 170.15

Link to comment

Hi,

 

Thanks for the great plugin, it is very handy.

 

I have a slight "bug\issue"

 

If I leave my server selection on auto I get poor speed results around 30Mbit/s Down and 3-4Mbit/s up using server (UBC Web Design (Ballarat)). But If I manually select a different server I get around my actual speed connections. I put this down to the server it is landing on, if I manually run speedtest-xml I see the same results.

 

If I run speedtest-cli I get a server (Telstra (Melbourne)) and speeds close to max (91-93 and 34).

However if I manually run speedtest-linux I get a different server (Melbourne (Telstra)) and speeds close to max.

 

So there seems to be a change in logic of server selection with speedtest-xml.

 

I am not too worried about gettting a MAX reading, I really run it too see if I get any peak speed drops, so manually selecting the server is OK.

 

But the issue I have then is that I seem to get more blank results back, than if I leave it on auto. I am not sure why it is failing (not much in the logs) but I know my connection is still active at the time of the test, I have other things running to detect dropouts. Is it possible that the -xml could have its server selection tweaked, or if when selecting manual you can have a backup server that is used if no result is returned? Or even the ability to black list some servers?

 

Blank Result Log:

Jun  3 00:00:01 TheBox speedtest: Internet bandwidth test started

Jun  3 00:00:12 TheBox speedtest: Host:

Jun  3 00:00:12 TheBox speedtest:

Jun  3 00:00:12 TheBox speedtest: Internet bandwidth test completed

 

Also, when I select "manual" on the settings page, I seem to get the same servers listed multiple times, I can't tell if they are different, name and distance is the same, maybe if it returned the server number as well so we could identify them?

 

Thanks again,

Wob

 

 

Edit:

Just some further info, with the plugin on auto it states it is using ping to pick a server, but the ping results are much higher than when I manually select a different server.

 

Auto

Jun  3 12:37:54 TheBox speedtest: Internet bandwidth test started

Jun  3 12:42:30 TheBox speedtest: Host: UBC Web Design (Ballarat) [151.03 km]

Jun  3 12:42:30 TheBox speedtest: Ping: 49.018 ms, Download: 30.90 Mbit/s, Upload: 3.65 Mbit/s

Jun  3 12:42:30 TheBox speedtest: Internet bandwidth test completed

 

Manual

Jun  3 13:02:17 TheBox speedtest: Internet bandwidth test started

Jun  3 13:02:46 TheBox speedtest: Host: AARNet (Melbourne) [225.60 km]

Jun  3 13:02:46 TheBox speedtest: Ping: 32.289 ms, Download: 91.89 Mbit/s, Upload: 35.79 Mbit/s

Jun  3 13:02:46 TheBox speedtest: Internet bandwidth test completed

 

Link to comment

Hi,

 

Thanks for the great plugin, it is very handy.

 

I have a slight "bug\issue"

 

If I leave my server selection on auto I get poor speed results around 30Mbit/s Down and 3-4Mbit/s up using server (UBC Web Design (Ballarat)). But If I manually select a different server I get around my actual speed connections. I put this down to the server it is landing on, if I manually run speedtest-xml I see the same results.

 

If I run speedtest-cli I get a server (Telstra (Melbourne)) and speeds close to max (91-93 and 34).

However if I manually run speedtest-linux I get a different server (Melbourne (Telstra)) and speeds close to max.

 

So there seems to be a change in logic of server selection with speedtest-xml.

 

I am not too worried about gettting a MAX reading, I really run it too see if I get any peak speed drops, so manually selecting the server is OK.

 

But the issue I have then is that I seem to get more blank results back, than if I leave it on auto. I am not sure why it is failing but I know my connection is still active at the time of the test, I have other things running to detect dropouts. Is it possible that the -xml could have its server selection tweaked, or if when selecting manual you can have a backup server that is used if no result is returned?

 

Thanks again,

Wob

 

I'll probably add an option to select which script to use since some seem to get different results. The speedtest-xml uses a beta version of speedtest-cli. So those can be easily interchangeable. You would just run speedtest-xml and it would use the original cli script or the beta depending on your settings choice. The speedtest-linux would require some code changes plus it doesn't have the speedtest image. So that may be a while.

 

Also I'll add a timeout option that should help with 0 results.

 

Edit: I can also add the server number to the list. It's what is stored in the config anyway.

 

Also I didn't write the underlying speedtest scripts so some of the functionality is beyond my control and ability to tweak.

Link to comment

Thanks dmacias,

 

Your solutions sound like they will do the trick, as I mentioned I mainly use it to test for peak speed dips, and it does a great job at that, just the blank results that are putting me out, so if there could be a check or retry for that it would be great.

 

Thanks,

Wob

Link to comment

Just playing a little more, I noticed on my box that the v0.3.1 script does exist, if I select that in the GUI I just get no result.

Thanks. I didn't realize --timeout wasn't an option for 0.3.1. That's the problem.

 

Edit: Updated so 0.3.1 works.

Link to comment
  • 2 weeks later...

Just noticed that I can't make the time change on the schedule stick. Make a change to the time and it will revert back to the old setting. I have completely removed the plugin and reinstalled, same issue,

Thanks. I get the same result. Must have forgot something when I fixed compatibility with dynamix trim plugin.

 

Edit: fixed now

Link to comment
  • 4 weeks later...

Hi,

 

I have started getting heaps of black results, more than actual results, i have timeout set to 60seconds, have tried both beta and 0.3.4 without much luck. changing the timeout to 30 seconds don't seem to stick for some reason, it just reverts to 60, i can set it to 20.

 

I have a set server, I have tried multiple servers without any luck.

 

Any ideas?

 

Thanks,

Wob

Link to comment

Hi,

 

I have started getting heaps of black results, more than actual results, i have timeout set to 60seconds, have tried both beta and 0.3.4 without much luck. changing the timeout to 30 seconds don't seem to stick for some reason, it just reverts to 60, i can set it to 20.

 

I have a set server, I have tried multiple servers without any luck.

 

Any ideas?

 

Thanks,

Wob

I've seen some blanks too. Been meaning to look into it. I'll have to add some more logging and see when it happens. I can change timeout and it stays but I'll look into that too.

Link to comment

My preferred server keeps on disappearing between runs.  Is there a way to just filter the server lists down to a zipcode or to manually enter in a server ID using the plugin?

 

I have not read enough to look into the CLI option.

Yes. On the settings page change Select Speed test Server from Auto to Manual. Then pick a server from the list and click Apply. The servers are sorted by nearest to your location.

Link to comment

Yes. On the settings page change Select Speed test Server from Auto to Manual. Then pick a server from the list and click Apply. The servers are sorted by nearest to your location.

 

That is what I am doing.  But my nearest server would disappear from that list.  Then if I pick another server and run a speed test and go back to the settings page, my nearest server would show up on the list again.  I would pick it, run a speed test and all would be fine.  Until I run another speed test and I get the message invalid server id.

 

If I run the speed test at the command line:

 

1. Can I specify an actual server id?

2. Can I force it to use sockets just like the current speedtest.net?

 

I have symmetrial gigabit and the plugin version is off by 100-200Mbps.

Link to comment

The list on the settings page dynamically retrieves the speedtest servers (that's why it takes a second to load). When you click apply it saves the server id to your speedtest.cfg. If you can recreated the invalid server id, let me know what is listed under SERVER from the config on your flash drive (/config/plugins/speedtest/speedtest.cfg) and what server your were trying to use.

 

There are four commands you can run from the command line. speedtest-cli, speedtest-beta, speedtest-xml and speedtest-linux. You can specify cli arguments on all but the xml since it uses speedtest.cfg. You can use "--server SERVER_ID"

 

The speedtest cli and speedtest linux were not written by me and don't include options for sockets. You can try speedtest-linux to see if you get better results. But the goal of these scripts and this plugin isn't to provide exact measurement comparable to running speedtest from a web browser but a consistent one that you can use to determine problems. The fact that they are written in python and different machines can run the code at different speeds can create inconsistencies between different computers. There are also known issues with gigabit speeds and not being able to show the full speed of the connection.

Link to comment

Thanks dmacias for the explanation.  BTW, excellent work on the plugin.  I have not been able to reproduce the invalid server error.  I noticed that if I toggle between the manual and auto setting for server on the settings page, my preferred server is sometime on the list and sometime not on the list.  My preferred server ID is 6285.

 

I have noticed that speedtest.net in the browser is really susceptible when you are testing gigabit connections.  At that bandwidth, things like browser and operating system compatibility starts to alter the results.  For example, on Windows 10, Chrome does not appears to be optimized to achieve true gigabit on the speed tests, whereas Microsoft Edge has no problem.  On Windows 7, Chrome yields the same results as IE.

 

I was hoping there was a CLI version of speedtest.net that can truly test gigabit connections using sockets to eliminate the browser and HTML from the equation.  All test sites that only uses HTML5 cannot correctly test the gigabit speed.

Link to comment

Hi dmacias,

 

Just wondering, would it be a hard task to port you plugin to a self contained docker? or other freestanding install, on say a Raspberry Pi? I want to have something that I can make a little more portable, and I haven't been able to find anything that looks as nice as your output (graph, etc)

 

Thanks,

Wob

Link to comment

Hi dmacias,

 

Just wondering, would it be a hard task to port you plugin to a self contained docker? or other freestanding install, on say a Raspberry Pi? I want to have something that I can make a little more portable, and I haven't been able to find anything that looks as nice as your output (graph, etc)

 

Thanks,

Wob

The graph page should be able to run on a webserver with some modification. It does rely on some dynamix functions and styling. But those could be eliminated or recreated. The only dependency this requires is python. The speedtest scripts should run on any linux distro. Just need to add a cron job, maybe change some paths to a user's home directory.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.