Howdy, Stranger!

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


Another terrible shitters
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.

Another terrible shitters

miumiu Member
edited February 2022 in Providers

Firstly fulfill within years servers leased through them (OneProvider) with hundreds of customers (and services bind with concrete IPs) and now after years of paying read such email:

Dear X Y,

Following an assets reassignment made by one of our partners in France, a comprehensive number of dedicated servers will be decommissioned on April 1st 2022.
We regret to inform you that some of your dedicated servers are concerned by this and will have to be replaced. A list of the affected services is included at the end of this email.

As part of our continued effort to provide you with the most competitive services, a list of exclusive offers has been prepared as replacement servers, within equivalent price range and specifications.
We have also built a unique self-check out panel, so that you may easily select your replacement servers and obtain them as seamlessly as possible.

And here is the good news: we will be providing you the first month of service for free, on all of your replacement servers.
You can access this special replacement panel here.

The following services will be affected:

xxxxxxxxxx

yyyyyyyyyyyy

zzzzzzzzzzzz

qqqqqqqqqqq

.......... etc

We understand the inconvenience this situation may cause and we want to reassure you that we will be providing direct assistance for the duration of this process if needed. Please do not hesitate to reply to this email to get in touch with your Account Manager.

  • Please note that replacement server stock availability may vary.
    ** If none of the options presented are satisfying, you may also request a full month of credit for each of your existing servers, which may then be applied to any of our services in the future.

IMO: renting servers from OP & online.net is great way to have later head in depression
One of my another GREAT mistake - build any infrastructure with them and start using them ever...

Thanked by 1darkimmortal

Comments

  • Damn. What server configurations are they shutting down?
    I have some i3s, L3426 with them but only got e-mail about price rise about two weeeks ago and nothing about EOL.

  • miumiu Member
    edited February 2022

    Seems all (or most of) France hosted

    All my FR are impacted (incl. relatively newer models - not only retro..)

    This sound me sooner that they want put over board customers who caught servers with low prices before years in special sales (BF and similar, my case..)

    I definitely end with this shitty provider, i need focus on progress in work, instead spend all time only for moving services with million troubles there then there then there around and do nothing more only kill all times with such shits..

    I spent at them more than 10k € all years
    this is their thank you!

    Thanked by 1rdes
  • @rdes said:
    Damn. What server configurations are they shutting down?
    I have some i3s, L3426 with them but only got e-mail about price rise about two weeeks ago and nothing about EOL.

    Additionally: maybe a bit possitive hope for u: prehistorical L3426 seems to be Dedimax not online.net

  • Also interesting: They cancel my 1240v3s, but offer as replacement 1230v2.. So great progress ahead.. But very evidently try force all affected customers to their new "dedicated EPYC instances" (yes i am happy jump from 1240v3 2x3TB SATA and 32RAM / 25€m on 2 vcore epyc with 125GB NVme for 15€m or 250GB for 29€, so tempting offer, especially where i have near to 3TB existing data on origin server, fantastic.. Or choose other server/ move and reconfig all there and after another half year expect another great new/email that your server is again EOL and u should choose another one for continue. So cute!)

  • So the offered equivalent isn't in the same price range? Tell your account manager and have them match it.

  • @TimboJones said:
    So the offered equivalent isn't in the same price range? Tell your account manager and have them match it.

    Thanks for your suggestion but I did probably biggest mistake when even began use them. Now i think i will do better when will rather move all at once and one time to more stable provider (who can guarantee less complications ans such cute surprises in future) than repeat this headache at OP after year or two again. Also their support was in most cases very useful (in style "dear customer we cannot, unfortunately this is not possible etc..")

  • @miu said:

    @TimboJones said:
    So the offered equivalent isn't in the same price range? Tell your account manager and have them match it.

    Thanks for your suggestion but I did probably biggest mistake when even began use them. Now i think i will do better when will rather move all at once and one time to more stable provider (who can guarantee less complications ans such cute surprises in future) than repeat this headache at OP after year or two again. Also their support was in most cases very useful (in style "dear customer we cannot, unfortunately this is not possible etc..")

    That's not realistic, really. Dedicated servers do need to be replaced every so often, at least within 5 years. And it's really your fault to hardcode IP addresses that you don't own! That's what DNS is for.

    If you're expecting to get long, uninterrupted life and pay low prices on old hardware, the problem is your expectations.

    Thanked by 3miu Daniel15 skorous
  • @TimboJones said:

    @miu said:

    @TimboJones said:
    So the offered equivalent isn't in the same price range? Tell your account manager and have them match it.

    Thanks for your suggestion but I did probably biggest mistake when even began use them. Now i think i will do better when will rather move all at once and one time to more stable provider (who can guarantee less complications ans such cute surprises in future) than repeat this headache at OP after year or two again. Also their support was in most cases very useful (in style "dear customer we cannot, unfortunately this is not possible etc..")

    That's not realistic, really. Dedicated servers do need to be replaced every so often, at least within 5 years. And it's really your fault to hardcode IP addresses that you don't own! That's what DNS is for.

    If you're expecting to get long, uninterrupted life and pay low prices on old hardware, the problem is your expectations.

    It is not so big problem rsync (or dd) partitions content to new machine. (My) problem is that one of server was NS1, and its IP is bind to ns1 parent record (glue IP) on more than 200 domains. (i know is better solution use NS servers of 1 "DNS domain" for all other domain as prevention of such failure and problem what is now, but not always possible because i do not own hosted client's domains and some of them ask for own ns with their domain name). All other is truth and i agree

  • miumiu Member
    edited February 2022

    @miu said:

    @rdes said:
    Damn. What server configurations are they shutting down?
    I have some i3s, L3426 with them but only got e-mail about price rise about two weeeks ago and nothing about EOL.

    Additionally: maybe a bit possitive hope for u: prehistorical L3426 seems to be Dedimax not online.net

    Oh sorry.. Now i realized that Dedimax is also only reseller of EOL online.net's servers and the same missery.. :-/
    https://www.dedimax.com/en/bare-metal-server/4

  • yoursunnyyoursunny Member, IPv6 Advocate

    @miu said:
    (My) problem is that one of server was NS1, and its IP is bind to ns1 parent record (glue IP) on more than 200 domains. (i know is better solution use NS servers of 1 "DNS domain" for all other domain as prevention of such failure and problem what is now, but not always possible because i do not own hosted client's domains and some of them ask for own ns with their domain name).

    Glue record is a not-good part of DNS.
    You can update a normal A or AAAA record and have it become effective in several minutes or hour, while updating a glue record could take days.

    I once read about a DNS provider that explicitly avoid using glue records.
    They have three DNS domains under .com, .org, and .dev, which belong to three different registries.
    When a customer signs up a domain, the provider assigns two nameservers that is not in the same registry as the customer's domain.
    This enables the provider to quickly change nameserver IP addresses when necessary.

    If I want to use "white label" nameservers, I would be creating them in two other domains that reflect my brand but under different registries.

    Thanked by 3miu idleparty FrankZ
  • @miu said:

    @TimboJones said:

    @miu said:

    @TimboJones said:
    So the offered equivalent isn't in the same price range? Tell your account manager and have them match it.

    Thanks for your suggestion but I did probably biggest mistake when even began use them. Now i think i will do better when will rather move all at once and one time to more stable provider (who can guarantee less complications ans such cute surprises in future) than repeat this headache at OP after year or two again. Also their support was in most cases very useful (in style "dear customer we cannot, unfortunately this is not possible etc..")

    That's not realistic, really. Dedicated servers do need to be replaced every so often, at least within 5 years. And it's really your fault to hardcode IP addresses that you don't own! That's what DNS is for.

    If you're expecting to get long, uninterrupted life and pay low prices on old hardware, the problem is your expectations.

    It is not so big problem rsync (or dd) partitions content to new machine. (My) problem is that one of server was NS1, and its IP is bind to ns1 parent record (glue IP) on more than 200 domains. (i know is better solution use NS servers of 1 "DNS domain" for all other domain as prevention of such failure and problem what is now, but not always possible because i do not own hosted client's domains and some of them ask for own ns with their domain name). All other is truth and i agree

    Ah shit, that does suck.

  • Good day and good bye

  • @KENTKING said:
    Good day and good bye

    Inflating post counts will lead to ban.

  • Oneprovider is just reseller of online.net , which has caused this shit , and this happened to large customers even expensive dedicated servers not just you !

  • This was announced 4 to 5 months ago to direct Online customers, and to be fair it was good whilst it lasted, most affected will have had several years of cheap service.

    But if OneProvider have only just let their customers know that is pretty poor.

  • tomazutomazu Member, Host Rep

    @miu said:
    It is not so big problem rsync (or dd) partitions content to new machine. (My) problem is that one of server was NS1, and its IP is bind to ns1 parent record (glue IP) on more than 200 domains. (i know is better solution use NS servers of 1 "DNS domain" for all other domain as prevention of such failure and problem what is now, but not always possible because i do not own hosted client's domains and some of them ask for own ns with their domain name). All other is truth and i agree

    ns2 would still work and probably you should add 3 more nameservers (ns3, ns4 and ns5) maybe on different gTLDs or ccTLDs, so that ns1 does not get hit that often and you make sure to turn the DNS service off on that ns1 IP a few days after starting the GLUE update.

    Am I missing something?

  • tomazutomazu Member, Host Rep
    edited February 2022

    @yoursunny said:

    @miu said:

    Glue record is a not-good part of DNS.

    it is a necessary part.. otherwise you will have a chicken and egg problem for at least one domain name on the entire internet. If you look at DNSSEC it gets even more complicated as even more interactions between parent and child zones are necessary. While you might argue that DNSSEC is over-engineered, I think the GLUE record is a valid, necessary and quite minimalistic solution for a real problem.

    Thanked by 1yoursunny
  • miumiu Member
    edited February 2022

    @tomazu said:
    ns2 would still work

    Hello

    Yes, sure will, only:

    1) in my experiences most requests (requesting www pages=domain's A records) will stuck and wait few-up to 5-10 second while NS1 should respond (if not responding then they go check another NS in order = ns2, ns3 and so on)
    Try this - disable DNS server where is running NS1 and when u will want load affected webpage your browser will wait few seconds bcs will wait if NS1 will respond or not (of course in case when u r using short TTL values (my case) for A records, not serveral hours or 1-2 days..)
    (i guess that DNS service always try check NS in alphabetic order, not round robin fashion)

    2) independently on the fact that services can work (with some initial delay/waiting for few sec caused when try firstly check NS1) still all affected domains using own ns should and need to be reconfigured (their NS1 IP), and at some registrars is changing ip for NS is real suffering/great fun..

    3) i had for example 3x6 TB SATA @ 1Gbit with near to 12 TB of utilized data in HW RAID5: they noticed me that will cancel this server and offer me possible replacements: such as 1TB SSD, 2x2TB SATA, and largest one 4x2 TB SATA with 100mbit, and in end prices rate a much more expensive as cancelled servers... Very cute.. How i should migrate >11 TB raid data on 2x3 or 4x2 SATA replacement? Absolutely they do not care about it - this is only my problem.. Another thing - zero effort try and offer to customers keep and move existing IPs to new servers (if they are interesting buy replacement ever at them still), this are 2 things what make me most disapointed ever on whole matter. (If they did offer really the same capacity config as replacement and option migrate and remain used IPs then i should be silent and not so disapointed from it)

    IMO very bad approach of provider.. Good Provider who really appreciate its customers will do not act so careless to them. What i think is that real reason is not any EOL of servers (paradoxically they offered as replacement also older CPU models as my cancelled were), real effort is drop over board customers with cheap large sata servers caught from previous special sale. But maybe i assume that they are simple total limited what online.net did and not able resolve this situation, so play rather that all is fine.

    All other is my mistake and stupidity (that i was not thinking in advance that they can do such something one day and was not ready for such situation), now i again got another strong lesson.

    So 3 lessons for me personally i learned on it:
    1) in future is should use for NS rather fail-over IPs (ovh, hetzner etc.), when i do not have own ip space yet
    2) rather order dedicated directly from their owners instead through any resellers on long-tails
    3) i should use good and willing providers who care for such things instead careless ignorants and "nice providers" like online.net, op and similar who want in first line get rid of old customers with cheap previous BF servers (maybe from reason of their bad economic situation.. i do not know this.. what is in the background at online.net)

    I already moved all on 4x8 SATA server, but i will good remember this my mistake and plan more sanely these things in future hopefully

  • darkimmortaldarkimmortal Member
    edited February 2022

    @tomazu said:

    @yoursunny said:

    @miu said:

    Glue record is a not-good part of DNS.

    it is a necessary part.. otherwise you will have a chicken and egg problem for at least one domain name on the entire internet. If you look at DNSSEC it gets even more complicated as even more interactions between parent and child zones are necessary. While you might argue that DNSSEC is over-engineered, I think the GLUE record is a valid, necessary and quite minimalistic solution for a real problem.

    The issue with GLUE is that it can only be updated via registrar, and in most cases this means wading through panels or even ticketing (eg namecheap ipv6)

    Sure it is needed to solve the initial chicken and egg problem, but once a GLUE record is set initially for a domain, I don’t see why the root servers can’t then use that to occasionally query for any changes to it

  • tomazutomazu Member, Host Rep
    edited February 2022

    @miu said:
    1) in my experiences most requests (requesting www pages=domain's A records) will stuck and wait few-up to 5-10 second while NS1 should respond (if not responding then they go check another NS in order = ns2, ns3 and so on)

    the caching DNS used will do this if there is no entry for the domain name in the cache; so it does not occur that often if TTL is not too low. But of course it is sub-optimal, I just tried to provide a solution to minimize the impact, not siding with the provider or anything.

    (i guess that DNS service always try check NS in alphabetic order, not round robin fashion)

    it depends on the client implementation, it was never completely "fair" distributed even when in the past GETADDRINFO was supposed to use round-robin (getaddrinfo still should deliver results round-robin), so modern clients do have a preferred order. But that applies in situation where e. g. ns3.serverclienti.com has more than one IPv4 and IPv6 address (RFC3484 had to implement changes due to IPv6).

    Looking at the stats of our geographically distributed nameservers I see only a slight bias towards those nameservers that are geographically closer to were most of our clients might be - but it is only marginal and it is not ns1 that gets the most queries, but ns3 (different geographic region). This probably is watered down a lot also due to nameservers like 1.1.1.1 of 8.8.8.8 that are present in many many data centers worldwide.

    That is why I suggest you add nameservers to minimize the problem.

    Sources:
    https://www.ietf.org/rfc/rfc3484.txt
    https://daniel.haxx.se/blog/2012/01/03/getaddrinfo-with-round-robin-dns-and-happy-eyeballs/

    @darkimmortal said:

    @tomazu said:

    @yoursunny said:

    @miu said:

    Glue record is a not-good part of DNS.

    it is a necessary part.. otherwise you will have a chicken and egg problem for at least one domain name on the entire internet. If you look at DNSSEC it gets even more complicated as even more interactions between parent and child zones are necessary. While you might argue that DNSSEC is over-engineered, I think the GLUE record is a valid, necessary and quite minimalistic solution for a real problem.

    The issue with GLUE is that it can only be updated via registrar, and in most cases this means wading through panels or even ticketing (eg namecheap ipv6)

    Sure it is needed to solve the initial chicken and egg problem, but once a GLUE record is set initially for a domain, I don’t see why the root servers can’t then use that to occasionally query for any changes to it

    well, the DNS system is hierarchical top -> down. Besides, you would need DNSSEC to implement such a mechanism in a secure way (UDP is spoofable, DNS cache poisoning etc.) and DNSSEC needs much more coordination between the parent and child zone anyways. So those are two possible reasons the root servers can't just use that.

    But maybe there could be some other mechanism, as described I just do not see GLUE as over-engineered as DNSSEC. The problem is with the Registrar that does not provide an API or other way to automatically update GLUE records if you ask me.

    Thanked by 1miu
  • Also got email today for Xeon E5-1410 v2.
    This servers also will be phased out by online.net.

    Thanked by 1emperor
  • @rdes said:
    Also got email today for Xeon E5-1410 v2.
    This servers also will be phased out by online.net.

    I had 5 FR servers, all affected=cancelled and as i said paradoxically they offered me as "newer replacement" also older servers (older and lower CPU types and downgraded config less ram, 100mbit instead 1gbit and similar nonsenses) as i had. Also old prehistoric xeon L3426 is still in their offer and available for purchase. So imo sure this is not about any EOL as they pretend, but real purpose and reason is cancelling cheap large sata servers from past/previous special sales from online.net managements geniuses. Ok, awesome approach, so i also cancelled both them from my provider list forever. I really don't care more about their other pleasant and cute surprises in the future.

  • It seems that they are replacing Online.net with Kimsufi/SYS/OVH, as their replacement configs are matching (e.g. one of their offers with 100Mbps exactly matches KS-16 with the only difference that OVH charges 19.99 and Oneprovider offer is 20.99, although the latter has 5-10% discount if paid quarterly/semi-annually/annually).

    Thanked by 2miu darkimmortal
Sign In or Register to comment.