Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


We operate 720 servers and we are happy to buy more. - Page 3
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.

We operate 720 servers and we are happy to buy more.

13

Comments

  • zGatozGato Member

    @Gulf said:

    @zGato said:
    I got it from there yeah, for 2300₸/m (roughly $5/m)
    They only accept CCs and at least Spain was blocked from making purchases until they manually lifted it for me.

    It is amazing if they provide such quality IPs nearly for free. Then no need for own tn.

    Performance and uptime is questionable, though...

  • jsgjsg Member, Resident Benchmarker
    edited October 7

    @nik said:
    Soo, just to understand the fancy name of IP triangulation correctly. Let's assume we are in the south of Germany with a local provider, or even DTAG (AS3320). You have a VM in Stuttgart or Munich with which you are doing your RTT testing and you would expect to have a low RTT since you are close to the VM. The issue is that EVERYTHING is backhauled through Frankfurt, so the RTT will be a lot higher and the user could be the same distance north of Frankfurt instead of south where the user is actually located. Am I missing something?

    Yep! The premise "time delay" (at least) roughly proportional to geo distance simply very often doesn't hold on the internet. In fact, I've seen more than my share of "funny" e.g. north-american carriers routing traffic from e.g. Prague to say Singapore via e.g. Seattle (and even worse). Same also within regions like you showed with e.g. DTAG and Germany.

    To actually do that sensibly one would need a model with nodes and vector edges. Simply pinging (or even tracerouting) doesn't do it well enough.

    That said I also often see the providers HQ location as that of some source IP when using IPinfo.io (or, for the sake of fairness, similar other services).

    Anyway, I don't care too much about IPinfo.io because for a non-commercial site (like mine) 20 or so free requests/day or else $50/mo is not attractive and I have to use really free services even though those probably are even less precise, but usually what I need is simple enough (like "from which country?" and "is it a VPN or proxy" is already kind of a luxury).

    @IPinfo.io

    I think you could extend your customer base quite a bit by offering an "low-end" paid option - but cheap, say $25 to max $50 per year for say 1000 requests per day or a total of 20000 per month.

    Thanked by 1niranjan
  • @Gulf said: It is amazing if they provide such quality IPs nearly for free. Then no need for own tn.

    I though about too but
    $5 for 1C1G
    $7,5 for 1C2G
    $12 for 2C3G
    or
    OWN $9 for 1C3G. Plus as zGato said OWN operates with EPYC.

    So, I would love to be cheapo but at that prices from telecom.kz still will pay for OWN.tn BOX.

  • alexhostalexhost Member, Patron Provider

    @Operable4829

    We have own Datacenter and Network in Republic of Moldova.
    Is possible to choose IP Ranges through Dedicated Servers / Colocation, wer also have BGP Services.

    You can check:
    https://alexhost.com

    ASN: AS200019
    Looking Glass: https://lg.alexhost.com
    Location: Moldova, Chisinau.

    If you are interested contact us through ticket or email and we will what we can do for long term :)

  • @varwww said:
    just curious, what tool do you use to manage all these servers?

    In terms of tools:

    • Tons of bash scripts
    • Airflow
    • BigQuery
    • Retool for internal apps and dashboards
    • Slack notifications

    The big thing is that there are a ton bash scripts that send health check metrics. If a server is not operational, the infrastructure team will manually investigate each server.

    It may sound like a lot when scaled to 720 servers, but we are quite stable :)

    Thanked by 1varwww
  • GulfGulf Member

    @SashkaPro said:
    OWN $9 for 1C3G. Plus as zGato said OWN operates with EPYC.

    They charge 22$ minimum.

  • @Gulf said:

    @SashkaPro said:
    OWN $9 for 1C3G. Plus as zGato said OWN operates with EPYC.

    They charge 22$ minimum.

    Nah. You are on LET. :)

  • GulfGulf Member
    edited October 7

    @SashkaPro said:
    So, I would love to be cheapo but at that prices from telecom.kz still will pay for OWN.tn BOX.

    Here is another rare location in KZ, reputation of netblocks is low for mail or proxies.
    So only for pings, like OP does.

    https://megahost.kz/vps/ - from 5$
    31.56.240.0/24 - KZ, Oral
    45.136.56.0/24 - KZ, Oral
    45.136.57.0/24 - KZ, Oral
    45.136.58.0/24 - KZ, Oral
    45.136.59.0/24 - KZ, Aktau

  • ShakibShakib Member, Patron Provider

    Not so accurate. :D

    ip:"*",
    city:"Manassas",
    region:"Virginia",
    country:"US",
    loc:"38.7510,-77.4753",
    org:"AS1239 Sprint",
    postal:"20108",
    timezone:"America/New_York",
    asn:Object,
    
    asn:"AS1239",
    name:"Sprint",
    domain:"sprint.net",
    route:"*",
    
        type:"isp",
    
    company:Object,
    
    name:"Sprint",
    domain:"sprint.pl",
    
        type:"isp",
    
    privacy:Object,
    
    vpn:false,
    proxy:false,
    tor:false,
    relay:false,
    hosting:false,
    service:"",
    
  • @Shakib

    Is this the information for you service's IP? Have you filled up the form for Hostcram?

  • @zGato said:
    Good luck finding it.

    How can I generate login banners like this?

  • HosteroidHosteroid Member, Patron Provider

    @Operable4829 sent you an message in private and also filled your form! :)

  • zGatozGato Member

    @jaimedelano said:
    How can I generate login banners like this?

    It's neofetch, you can install it via apt. It's not a "log in" banner but I just typed it in.

  • GulfGulf Member

    About the database itself. I checked every IP, subnet and ASN, and found a lot of mistakes.

    Like this ASN is ISP, although it is hosting, even the name of the company is related to hosting
    https://ipinfo.io/AS13739

    Any random example, this one is also hosting
    https://ipinfo.io/AS37349

  • GulfGulf Member

    If need rare locations, send the country / city name.

    I test dozens every week.

    But there are problems, many hosters, outside this forum, are bad :D
    Especially in these locations.

    Ignoring support tickets is ok.
    Refund is not refund. Some even require crazy conditions for refund.
    Rudeness is common.

  • zGatozGato Member

    @Gulf said:
    If need rare locations, send the country / city name.

    I test dozens every week.

    But there are problems, many hosters, outside this forum, are bad :D
    Especially in these locations.

    Ignoring support tickets is ok.
    Refund is not refund. Some even require crazy conditions for refund.
    Rudeness is common.

    You can get Nairobi, Kenya from Selectel :)

  • GulfGulf Member

    @zGato said:
    You can get Nairobi, Kenya from Selectel :)

    It is amazing, but they rent it at https://servercore.com/

    There are low number of providers there, here you can buy vps / dedicated, but need to contact them
    https://afriqnetworks.co.ke/contact.php

  • Hey guys, sorry for going completely silent. I am trying to catch up

    So far, we have only bought one server which was based on network metrics. Thanks, @zGato . Even though we had servers in that location, the network metrics showed the importance of the server.

    I reached out to the some vendors @Gulf mentioned. Even though they did not have self-serve, they responded to my email and hopefully, we will be able to get a deal going.

    Again, I recommend you guys post affiliate links in the forms. Like it is the least we can do here. Thanks!

    Thanked by 1zGato
  • GulfGulf Member

    Great, try more rare locations :D

    Tanzania
    https://aptus.co.tz/vps.html

    Bolivia
    https://www.confiared.com/

    Honduras
    https://intrapres.com/ - need to contact them (they may be host here https://ipinfo.io/AS273189)
    https://www.hyve.com/ - they also host in AS273189

    Mauritius
    https://www.kaldera.mu/

    Curacao - no idea, why russian sell here
    https://sincere.cloud/
    gbnhost.com - https://ipinfo.io/AS35112
    cloudx.ru

  • GulfGulf Member
    • more

    Aruba + Curacao. May try:
    https://www.netprogroup.com/ - need to contact
    https://www.setar.aw/business/ - here they listed hosting

    Ghana + Ivory Cost
    https://mainone.net/products-and-services/ - need to contact

    Trinidad and Tobago
    [email protected] - email only

    Aland Islands
    https://www.connect.ax/ - try to contact

    New Caledonia + Papua New Guinea
    https://www.mynet.nc/ - also need to contact
    https://www.mynet.at/ - the same company

    Cayman Islands
    https://www.c3.ky/ - need to contact
    https://www.salt.ky/

    Monaco
    https://www.monacocloud.mc/notre-offre

    Gambia
    https://unique.gm/

    --
    Will send more. May also send rare cities. Send names.

  • zGatozGato Member
    edited October 12

    Well, like I said, you send a correction, it gets accepted, few days after it's back lmao

    Another example

    And another

    But for others you don't, hmmm
    (screenshot from July)

    It's quite sad, honestly. You're making it too blatant.

    Thanked by 1AlbaHost
  • GulfGulf Member
    edited October 12

    Yes, many mistakes :( Not sure if this is an appropriate thread to report.

    Example 129.185.24.1

    It is located in France, why did you place it in the US?

    https://ipinfo.io/129.185.24.1

    Trace route ends at 92.43.248.90

    And there are thousands errors like that...

  • @jaimedelano said: How can I generate login banners like this?

    https://cloudybarz.com/custom-motd-in-linux-debian/

  • GulfGulf Member

    Any random check...

    This one is Hong Kong, not Singapore
    https://ipinfo.io/223.120.23.1

  • zGatozGato Member
    edited October 12

    @Gulf said:
    Yes, many mistakes :( Not sure if this is an appropriate thread to report.

    Example 129.185.24.1

    It is located in France, why did you place it in the US?

    https://ipinfo.io/129.185.24.1

    Trace route ends at 92.43.248.90

    And there are thousands errors like that...

    It was actually in France previously but changed for some reason?

    I'm not even mad at them for having a few addresses with wrong accuracy. My home IP is located like 100 km away from where I am, whilst their competitors (e.g. MaxMind) are within a 5 km range.

    It's just very coincidental that most big VPN companies out there mysteriously don't have any geographical checks, and they just believe what they tell them. And it's not like they can't ping the address to check where it's really located, cause you send them a correction, they approve it and display that banner, and days after it's back to what it was. This does not happen to any other IPs I've sent corrections about, only to SurfShark/NordVPN IPs.

    Look, another example of Nord!


    Just remove their stupid geofeed from your system already
    https://geofeeds.cyberzonehub.com/geofeed.csv

  • GulfGulf Member
    edited October 13

    @zGato said:

    Try 198.252.105.0 in your checkers. You can remove these who show Canada :D

    Yes, very hard to find one comprehensive database.
    Sadly, ipinfo does not provide cities.

    PS try also 172.96.188.0
    Looks like only ipinfo shows it properly.

    PPS also, problem in this IP in ipinfo 212.94.84.77

  • GulfGulf Member
    edited October 13

    172.96.188.1

    Here is an interesting example, you can investigate by pings :D

    Hoster states it is located in New York and ipinfo also shows New York City
    https://ipinfo.io/172.96.188.1

    However, the advanced paid Maxmind database shows New Jersey, Secaucus.

    The facility is SingleHop. You can also find that SingleHop has a dc in Secaucus.

    So who lies here, ipinfo + hoster or maxmind advanced database? :D

  • zGatozGato Member

    @Gulf said:
    172.96.188.1

    Here is an interesting example, you can investigate by pings :D

    Hoster states it is located in New York and ipinfo also shows New York City
    https://ipinfo.io/172.96.188.1

    However, the advanced paid Maxmind database shows New Jersey, Secaucus.

    The facility is SingleHop. You can also find that SingleHop has a dc in Secaucus.

    So who lies here, ipinfo + hoster or maxmind advanced database? :D

    And it's not like they don't know which IPs are pingable or not, they just decide not to check it.
    Example of NordVPN "Bahamas" location
    https://ipinfo.io/tools/pingable-ips

    and yet still decide to not ping and check the real location

  • GulfGulf Member

    @zGato said:

    Thank you I check the site you use.

    IPGeolocation.io passed all ips above, except, 223.120.23.1 (detected SG instead HK)
    IPregistry.co passed all, except, 212.94.84.77 (detected city Vantaa, instead of Kuopio, need to check Finland geography, maybe this is related)

    Other services failed test.

  • Operable4829Operable4829 Member
    edited October 14

    Hey guys, @zGato and @Gulf

    Sorry again for not responding sonner. I got busy over the last week. I really appreciate you guys, investigating our data.

    In almost all of the IPs you flagged, there are location hints indicating where the IP address is actually located. The reason the location was not shown was due to an internal bug. The issue is about trusting the geofeed.

    As one of our engineers said:

    "We do use RTT measurements to validate geofeeds, however in some instances RTTs legitimately mismatch the user location (e.g. CG-NAT) in which case we override RTTs measurements with the geofeed. However it may happen that we have trusted some feeds that shouldn't be. We're constantly improving our processes to find these feeds and stop trusting them."

    We do have a running project to identify the ASNs with intentional misleading geofeeds and one of our engineers is leading it.

    I have also created reports for each of the IPs you have flagged diligently.

    Majority of location hints point to Miami. Overwritten by geofeed priority.

    Location hints point to Marseillie, FR. Overwritten by geofeed.

    Location hints point to Singapore. Overwritten by geofeed.

    Location hints point to France, but one data source which is not geofeed had a slight edge of approximately 8% higher. For that location, it went to the US. We are investigating this issue.

    If you ping this IP address from multi ping service like ping.sx you will see a mix of HK and SG. In this type of sitaution we have to pick one. As we described so far, we have to pick one, we ended up picking SG.

    Location Loss Sent Last Avg Best Wrst StDev
    Singapore 0 10 2.3 2.34 2.16 2.42 0.08
    Hong Kong 0 10 39.1 38.28 38 39.1 0.31
    Hong Kong 0 10 44.6 44.56 44.38 44.8 0.14
    Singapore 1 10 48.42 43.69 48.32 48.75 4.8

    This one is probably in Columbia. Again geofeed overwrite.

    We are pretty confident it is in Dallas, TX, US.

    New York City, US had a literal 2% edge over Secaucus, US

    Pretty sure it is in Finland.


    It's just very coincidental that most big VPN companies out there mysteriously don't have any geographical checks, and they just believe what they tell them. And it's not like they can't ping the address to check where it's really located, cause you send them a correction, they approve it and display that banner, and days after it's back to what it was. This does not happen to any other IPs I've sent corrections about, only to SurfShark/NordVPN IPs.

    I understand that. Please be a bit patient. This seems to be a bug, and our engineering team is actively investigating this issue.

    I understand that reducing the priority metrics for the geofeed seems like an immediate solution, but the challenge is that the engineering team needs to evaluate how to pick the data source when there is conflicting location hints or in the case where an IP address cannot be actively measured. The goal is to provide reliable and accurate data, so, we can not risk pushing a hotfix and break something in the pipeline. We cannot compromise on the reliability aspect of the service, so the engineering team needs to thoroughly investigate, vet, and test solutions before deploying them.

    We appreciate the corrections you have submitted but it seems like the issue is related to a known issue and our engineers will focus on addressing on the root cause first.

    I will, as always, keep you posted on the product updates. I am trying to catch up with one of the server recommendations you made so far. Again, thank you very much for such great feedback and support you have shown me so far.

Sign In or Register to comment.