CA3LE

Israel would be a great test location. Show me the great, cheap VPS hosts with lots of bandwidth to offer.

Show me great VPS hosts in those areas that I can show are great and I'll add them ON REQUEST. Any time. Please help me in finding the best hosts in ALL of your areas. Better hosts = a better speed test. Better hosts remove the variable on MY end so I will not host unless they prove their weight. Trust me, ask people around here. I throw servers up for people all the time, I just need a VIABLE host in the area that tests out properly.

Glad I released two days early to catch this stuff. The real release is 4/15/19 .... 4+15=19 = v19 We're here now, thank you guys.

I really love how you guys hold me to a standard. I can't deviate from what you expect. And that's awesome.

Nice. You rule for getting back to me so quickly.

the reason this is an issue now and not before is because the front end shares more code with the actual speed test. That's a frontend speed improvement that obviously for this purpose is NOT an improvement. But for loading the download test page that you use to select your test... that's a good hack. --- those are actually (and have always been) the same program, even though they became separated from each other (away from the main server) when TMN expanded to multiple servers... long ago. Things that may not have affected the test-side... let's just say I need to be more careful. You guys are awesome for catching all that you do. I truly could not build TMN without you... do you know how much testing I've done!?! Some do.

Maybe that z-lib compression fix helped your situation as well. Please let me know. -D

I only query you for more information in case I end up having a hard time finding the issue. Actually, what you posted was more than sufficient in this case. I looked into it and was able to replicate the issue using a virtual machine. My own machines (and I assume my beta testers) won't display this behavior for some reason. I did see this in virtual machine before but I thought maybe it was a fluke, guess I should have dug deeper. It was an easy fix. The header was setting "ini_set("zlib.output_compression", 4096);" --- I was totally replicating your results so I already know this is fixed. DUDE, THANK YOU! Great catch!

Share your results please, I'm not seeing anything by your username or computer id (associated with the IP you're porting from)

What are the specs on your computer?

I'll look more closely, we hadn't seen anything like this in our beta testing. I've been seeing a slightly higher upload speed, but it's due to an improved process. The download test under everything you see... is nearly identical. The processes and functions are largely unchanged. I'll look much closer at the results. I make mistakes and appreciate that you hold me to a high standard, helps me find and learn from those mistakes. Food for thought, when I make major updates I do them in stages. First stage is what you see... visually. Then comes the real programming. I do deep programming during the visual stage but for the most part, I have to integrate the design into the program... (TMN is not built on a CMS by the way, I like it that way). the program you're using right now (the new version) is the exact same program. They share the exact same code base... its branched off quite a bit at this point but the core is basically the same. It's really hard to explain. [page loads] [checks if you have a cookie set telling me you want to use one version or another] [keeps loading the program] [encounters an instance where the code branches into two versions (or more)] [loads the proper version data] [keeps loading the program] [encounters an instance where the code branches into two versions] [loads the proper version data] [keeps loading the program] etc. the base is the same, a few elements are altered. ... Wanted to help you get a feeling for how it works under the hood. If you understand that this isn't at all an entirely new program... it may help you feel that it's the same TMN (I feel) you've grown to love. The crazier updates come later.

Sweet. Thank you SO MUCH for being responsive and helping me resolve this issue. Very happy to know this resolved. The root cause was that the servers you had selected (the UK group) were not performing their automatic update checks properly. They hadn't received their updates. It's all good because it prompted me to fix a TON of other issues related to that. One major one was blocking the process in Firefox. Resolved. -- didn't used to do until fairly recent. Browsers update things and it will break behavior and you have to move with it. That server hadn't received its updates from the master (there's one master mirror server that feeds updates to the others), the servers you were selected on were clones so the error had been repeated a few times. Only way to be sure was to check them all. I went through the entire server stack, they're all fully updated with the latest everything and are set to properly maintain themselves into the future. Also styled the auto test with the new styles. Still not done everywhere. The crontab was just empty on those servers. ? /sigh

go to your command prompt / terminal and ping testmy.net it will return the CDN IP that resolves for your location.

Thanks Gabe!

Are you sure you don't have another browser window open that's running another auto test? It's iterating properly on my end properly but it could do what you're saying if there were two of them open at the same time, in the same browser.

Can you give it another go please?

This gives me a better idea of where to look... it's not anything inside the actual v19 auto test program. Pretty sure now that it's in the frontend code somewhere. Give me a little time to search. I had a variable that needed to be disabled to fully put everything on the backend into the legacy version. Long story short, can you run that again and make sure it still does the same thing. If it still does, then I can be pretty certain it's in the frontend client code. Narrows my search down. thank you!

I saw it do that in development. @Sean was pointing it out to me months ago and I thought it was resolved. I did see the issue again just now, having a hard time seeing why it's intermittent like that. I told the program to load the old auto test source code but to have everything else run the new version. When you get a chance, either @Drip LeBuk - @Michael-Buckley or anyone else who's seen this issue, please test this again and let me know what you see now. Thank you

Thank you... and no, not right now. I'll need help testing it later though. Just use that link at the bottom like you did before to run on the old version for now. I don't see this in any of my testing yet but I'm sure there's good explanation.

Test completed right as I submitted that last post. Hummm, I'll keep looking.

I loaded up Windows 10 with Chrome 73 and couldn't reproduce the issue. I'll keep looking. It's running a test right now.

I'll give it a try on your exact setup and get back to you soon.

That's odd, I just tested and I'm seeing it. Maybe try clearing your cache. Please send me a screenshot of what you're seeing on your end.

Thank you for the suggestions, I appreciate the feedback and will work on those ideas. ?

There may be poor routing from your host to mine or vice versa. These types of problems are outside of your control and mine. Sometimes it can affect larger portions of your traffic other times it can be more specific to certain data centers. As long as you feel that your running normally everywhere else, I wouldn't be at all concerned. Glad that the mirrors helped clear things up. I see you're using the beta, do you like it better?