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
Being all suspicious I sent 5000 DNS "random" hostname requests to a domain hosted on Rage4 with a C# code looking something like this:
for (int i = 0; i < 5000; i++)IPHostEntry = Dns.GetHostEntry($"test{i}.bspeed.co");
There was an A record *.bspeed.co pointing to some IP.
Suddenly my EU query count jumped from ~130 a day to 58.000.
I'm running another test just to make sure... Will edit this post as it goes
EDIT: Yeah, my second test confirmed this. The new count jumped to 114k. Can I have my 50€ back now?
.
Then I sent 1000-2000 requests (I cancelled in the middle and I can't remember the actual number) to another domain hosted somewhere else and the query count was 1300....
Time to move away I guess. I paid my last 50€ invoice and hopefully I'll find someone who doesn't have some weird shady pricing. I wish I had done this test earlier.
>
Before you move away, wait for an answer from @gbshouse.
The word "scam" gets thrown around quite often here, but it might be appropriate in this case.
I'm not accusing anyone of scam, I think this is just really bad design and something is wrong somwhere. I think it has something to do with cache and how they count requests. But for me as a customer, I guess this might be called scam. Especially since this was adressed many times and they said it's perfectly correct.
If it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck.
I'm just upset I spent 400€ with them. Meh..
They always counted significantly more. That being said, I have one of those prometeus accounts so I don't care.
Who are you going to use now?
Their count was always off. 1 million queries on some domains per month, prompting me to get the $19 euro whatever unlimited plan.
Decided to get DNSMadeEasy / proper Anycast, and 35 domains (got 10 addon) don't surpass 1.2 million per month.
It's either bullshit or poor coding, but it's present.
It is very sad, every review of Rage4 I see mentions the abnormal counts.
If you ask me it's probably counting DNS questions as queries and counting their internal queries as paid queries or something weird like this. Either that or it's straight bullshit.
For example if they get an ANY type DNS query they would count it as both A, CNAME, MX etc..
I built an in-house DNS system for BunnyCDN, but for my personal stuff I'm using Amazon Route 53 simply becaue they are actually pay as you go, not 1 query over a million, one € more and quite popular and simple to use really.
What is your TTL?
It was 5 seconds, but that doesn't really matter since the requests were done to "random" hostnames.
I agree with the OP. We sent some customers their way and later found DNSMadeEasy that does failover without a third party monitor, seems slightly faster and costs a whole lot less. They're good - but on the expensive side.
@BunnySpeed - we are always clear regarding our pricing and why the usage count in our system is different. I've already explained above how our caching works. Running query usage tests, especially via public DNS infrastructure is not the best idea, especially when using Google DNS or similar. In case of Google they never query just one closest server, instead they use 3-5 different locations and each query is sent from different IP (you can check all their ranges on their support page) which literally amplifies the traffic. Also if you have sent the query like this "test{i}.bspeed.co" without proper formatting (like String.Format etc.), the "i" was not replaced and you were charged for "bad" queries.
It's more than 3 years now since we've launched Rage4 DNS. We provide our services to all kinds of business, also providing platform-as-a-service to third party providers.
Our billing was always transparent, beside standard pay-as-you-go we have whole set of flat-fee based packages and if you need something custom we are always there.
If anyone has any questions regarding how our stuff works it's enough to open support ticket.
The formatting is fine, it's the new C# syntax.
Why are you blaming this on google DNS? I tested the same code on our DNS system and it logged approx. one DNS request for one call of DNS query while yours logged 50k requests for 5k requests.
Since you blamed Google DNS I decided to run another test sending simple A type DNS queries DIRECTLY to your nameserver. I sent 10k queries and the counter jumped from 275k to 335k. Running a second test just to make sure.
EDIT: Once again, same thing. At this point I can't even blame any kind of ANY requests or caching since I'm clearly sending "banana.bspeed.co" A type DNS queries one after another. Jumped from 99 to 37k (60k after a few more minutes). Care to explain?
This is the code I used for the test (I reused some of it from some other testing jobs for other stuff, that's why it's a bit messy). The upper commented out code was used to capture the packet and the bottom part was used to send it. The packet was generated by a testing tool from http://simpledns.com/ (And please don't blame the tool)
You were comfronted with this a multitude of times yet you always say that the counter is perfectly fine and blame resolvers, caching, etc. Well, I don't believe you anymore unless you explain this to me. It seems that your queries are inflated up to 10 times. My guess would be 5 times + the original queries.
Everyone who's used Rage4 here has the same impossible experience with their request counts and he's never been able to give a satisfactory answer.
The only conclusion I can come to is either incompetence or malice. Hopefully it's the former and they simply don't know what they're doing, but their refusal to even look into things makes the latter seem equally as likely. I certainly hope they're not running a pricing scam, but if it looks like a duck, swims like a duck, and quacks like a duck...
Since I have a partners account, and can perform tests without it costing me any money I did some testing. Some may see this as a conflict of interest so please feel free to repeat testing with the published code, or critique the code as necessary.
All domains used are not registered, this ensures a background request rate of 0r/s. The code use for testing has been uploaded to Github (https://github.com/splitice/Rage4Test). During development of this code all testing was done on different domains, the only requests sent to these domains are from a single execution of the testing application.
Test 1:
Rage 4 DNS Configuration:
Test: 10,000 requests to blah.test-r4-1.com
Expected Usage: <10,000. If caching working as I understand it possibly then ~$NUM_NODES so lets say <100
Resulting Usage:
315 (http://img.x4b.org/04-05-2016/14_30_29.png)NOTE: Test repeated due to issue in initial run.
Test 2
Rage 4 DNS Configuration:
Test: 10,000 requests to blah-$i.test-r4-2.com
Expected Usage: ~10,000
Resulting Usage: 60000 (http://img.x4b.org/04-05-2016/14_40_54.png)
Conclusion
Test 1 appears to return the correct result.
Test 2 appears to be incorrect / inflated by a factor of exactly 6.
I'll check back later today and make sure the usage does not increase (they are not real domains, they shouldn't have any requests). I havent seen any stats creep yet, so it appears to only be an issue where requests are counted multiple times.
I have had alot to do with Piotr over the years, and I dont for a minute believe that he would purposefully skew usage results. If there is any error, I expect to be the result of a bug, or some unexplained variable. I encourage everyone affected to behave civilly, rather than yelling SCAM, there is enough of that on this forum, and it accomplishes little.
I will point out I performed this test from the US East Coast, and they appear to be hitting EU nodes, or atleast being counted as queries nodes. I'll cut Rage4 some slack since I am querying direct, but I do agree, perhaps the Anycast routes could do with some tweaking and testing (or perhaps thats just an issue with the usage display).
I moved to cloudns and never look back.
I'm glad I saw this post. I've had 3EUR/mo for the past few months but will be looking at backup options.
I'm glad you wrote some proper code, I just threw something together, however your test is incorrect. That's why I was avoiding any DNS libraries and clients. Your DNS client is the one doing caching here in Test 1, not Rage4.
See (https://github.com/ghuntley/Heijden.Dns/blob/master/src/Heijden.Dns/Resolver.cs). You should set UseCache to false.
When I was referring to caching I was merely implying that they might be charging additional queries if the query is not in RAM and they fetch it from another node or some weird thing like that, but that should only increase the count for maybe 5 queries max in a total test of 10k, If the request goes to their server it should always be counted. I think you misunderstood how caching works. It would be different if you were hitting Google's DNS for example that could cache things.
This is all starting to look quite sad.
As you noticed I don't just throw the word scammer around lightly, but @gbshouse this means I most likely paid over 330€ too much for your queries. Makes me feel all warm and fuzzy with love.
I will set UseCache to false and repeat, I didnt even notice that property. Thats why we open source boys
And the cache I refer to is this one:
I interpret that to mean, queries that result in a SQL query are counted. Those that are handled by the PDNS packet cache are not.
I think at this point we should continue this in a new thread: https://www.lowendtalk.com/discussion/82652/rage4-dns-counts-inflated-queries
Hopefully I didn't forget to add anything. If I find something I missed I'll just add it to my post.
@BunnySpeed FYI Test 1 repeated and got a result of 15.
@BunnySpeed - I'm not blaming anyone, jest telling you what I can see sometimes from this side of fence.
I'm happy to sign the NDA with @SplitIce or any other trusted user and let him or her inspect our code.
I also had no intentions of attacking you, in fact I always recommended you to everyone. But CLEARLY there's something wrong yet you don't even admit it seems abnormal. Yo're just saying your code is nice and fine. It's making me all raged up
Blaming your network design also isn't an excuse. If a user makes one request, I want to be charged for one request.
@BunnySpeed his response reads pretty ok to me.
You should respect that as a provider his options for defending himself are limited. He is (currently) the only person who can see the code, and state that (to the best of his knowledge) it is accounting accurately. Anything else is speculation.
Now, he has presented a way forward, for someone (assuredly trusted & respected individual by both the community and GBSHouse) to do an inspection under NDA.
I am not going to put my name forward unless LET asks me to, as I have a partners account and engage Rage4 for business purposes. That is I have a potential conflict of interest (even I know I will never act on it). That being said, if I am requested I would volunteer my time.
This "duck test" is absurd. It hints that it is based on observation but it is not.
You can invalidate it easily:
Your child could dress up like a duck, swim like a duck, and quack like a duck, but that does not make him/her a duck. You have to be stupid to think that your child probably is a duck.
Just look. Observe. Don't make assumptions.
Facts:
I don't see any scam here. How many times a coder puts his code into production ENV just to realize that: "omg there is a huge bug in the system!"
Maybe he did not use tests. Maybe the tests are fucked up. Maybe @gbshouse is evil. Maybe because they are Italian it is connected to the Mafia. Mafia wants to take over the world by billing much higher usage. But these are just pure assumptions... no observation involved.
If you see the half of a car that it is not hid by the wall, it does not mean that there is a whole car behind that wall. Yes you can assume that "well it must be a whole car... because thats what I always see." But that does not make it a fact, until you observe it.
So please, observe the obvious.
and your child is not a duck...
Sorry for this long post, but it is happing to much times when "scam" is used on LET based on assumptions, opinions.
That's exactly why I'm doing further tests, and in the topic I created exactly for this discussion I think I developed a pretty solid idea why this might be happening. I wanted this one to be a positive review, that's why I created a new one. I also never said they scammed me.