Howdy, Stranger!

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

Advertise on LowEndTalk.com
EDIS reliability issues
New on LowEndTalk? Please read our 'Community Rules' by clicking on it in the right menu!

EDIS reliability issues

jon617jon617 Member
edited April 2013 in Reviews

Am I the only person having reliability issues with EDIS recently? I have 12 KVM VPSs with them, just checked all logs, and I have a total of 10 server reboots in the last 4 weeks. That averages out to about 2.5 of my 12 VPSs reboot each week. Even if I had only one VPS with them, the reboot frequency would average out to one reboot every 12 weeks for each VPS. Maybe I have high standards, but 3 month average uptime seems a bit unimpressive.

Sometimes when my VPSs go down, they don't reboot by themselves. I have to go into the Edis control panel and issue a hard reset.

One of my VPSs has been down for 5 hours. Tried hard reset and forcing boot from the panel, but no luck. Sent them an email 5 hours ago and waiting for response. Granted it is 6:30am Sunday in Austria now and this issue is happening in the middle of the night, but I would hope a company the size of Edis would have someone watching for "server is down" emails in the middle of the night.

Are these issues just me, or are others having similar experiences?

«1

Comments

  • Did you try asking edis about this issue before posting on let?

  • I never had a KVM from them, just vserver and is doing great.
    This kind of instability for kvm is concerning, tho, in my view should be as stable as Xen.
    OTOH, for the price, a reboot in 12 weeks is not the end of the world, not coming back after, tho, that's at least bad.

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • @skirtTight said: Did you try asking edis about this issue before posting on let?

    Should I have done more than sending them email?

  • @jon617 said: Should I have done more than sending them email?

    Oh yes, ofcourse, wait for a reply.

  • jon617jon617 Member
    edited April 2013

    @Maounique said: OTOH, for the price, a reboot in 12 weeks is not the end of the world, not coming back after, tho, that's at least bad.

    I agree, an average 3 month uptime per VPS is not that bad, but it's not great either. Actually, I should have checked this first...but I have 3 KVM VPSs in the USA, actually they also seem to average about 2-3 months between reboots, so I guess my issue with Edis is probably not abnormal for a KVM. So let me clarify after thought...I only notice this because sometimes the Edis VPSs go down for a while (more than a few minutes at a time), and sometimes don't come back online without manual intervention. I'm not ripping them, service has generally been great especially for the price, just wondering if others have seen similar KVMs not coming back online.

    @Maounique said: This kind of instability for kvm is concerning, tho, in my view should be as stable as Xen.

    In theory, I too would think KVM would be more stable. I have one Xen VPS elsewhere which had a year of uptime before I rebooted it to update the kernel.

  • <3 EDIS

    .

    Hello, my name is Cloromorpho and i am a LEB addict.

  • My Edis KVM reboots every few weeks as well, but it comes right back up. I don't plan on renewing though.

  • maybe @William can help?

    xD

    Hello, my name is Cloromorpho and i am a LEB addict.

  • 10 different servers rebooted in a month sounds like they are rolling some updates and rebooting the nodes after that? Anyway, probably only they can tell.

    -

  • I've seen the same behavior with one I have with them as well.

    Uptime monitoring for the masses. NodePing

  • WilliamWilliam Member, Provider

    Linux bridges have the bad behaviour to lock-up at DDoS... so you can guess that the cause of most problems is...

    (And thus the overall downtime is higher with us as we allow IRC and pretty much anything else in most locations, which sometimes can (and do) cause DDoS attacks)

  • Now I know why many providers disallow IRC.

  • krokro Member

    I notice the panel doing qwerky shit as well

  • IshaqIshaq Member, Provider
    edited April 2013

    @Voss said: Perhaps it is time to re-consider that policy?

    Perhaps not, I run 2 IRCds with them and they work fine, they don't attract unwanted attention.

    [BudgetNode] DDoS Protected. 7 Locations (US/EU). Check out our latest offer!
  • Disallow IRC on business lines, allow on budget but take down customers that get involved in attacks, either receiving or sending. They want to do s**t, fine, get a protected service.

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • rm_rm_ Member

    From my experience with a dozen of providers all allowing IRC (otherwise I don't buy), IRC is not the problem.

  • WilliamWilliam Member, Provider

    @rm_ said: From my experience with a dozen of providers all allowing IRC (otherwise I don't buy), IRC is not the problem.

    For us it is.
    DDoS attacks are currently (did some portscans and had customers tell me what was attacked) spreaded around this:
    25% IRC servers
    25% VPN/Proxy providers
    25% Camfrog
    20% Unknown/80
    5% DNS/UDP

    @Maounique said: Disallow IRC on business lines

    huh? We don't have any separation - we only have one sort of packages and that will likely not be changed.

  • DomainBopDomainBop Member
    edited April 2013

    Linux bridges have the bad behaviour to lock-up at DDoS... so you can guess that the cause of most problems is...

    The only reboot I've had on any of my 6 Edis KVMs in the past several months was Iceland this week due to a DDoS attack on someone on the node but the downtime was minimal. By contrast the OVH Roubaix VPS (which will not be renewed) I'm testing rebooted twice during the past week. and the ChicagoVPS Atlanta VPS I canceled this week (and switched to RamNode which is much more stable) had rebooted twice in the week before I canceled it.

    Are these issues just me, or are others having similar experiences?

    Best way to find out if it's just you or the entire node is to check the status page
    http://status.edis.at/

    this kind of instability for kvm is concerning, tho, in my view should be as stable as Xen.

    The longest uptimes and most stable (excluding HA cloud solutions like CloudVPS) of any of my VPS's aren't KVM or Xen boxes...it's the openvz boxes at IntoVPS where 200+ or even 300+ days of uptime are the norm. IntoVPS's openvz uptime magic is probably due in part to the fact that their pricing makes them unattractive to DDos targets/abusers like HF kiddies, VPN and IRC users, etc.

  • WilliamWilliam Member, Provider

    OpenVZ is rock stable, also under DDoS or packetstorms - as it uses no bridge :)
    I'm not sure about Xen but i'd guess they use an internal bridge which is better engineered than the normal Linux one.
    KVM is, well, KVM.

    The only really stable bare metal virtualization seems to be VMWare, which runs stable under any amount of packets/mbit.

  • @DomainBop said: The only reboot I've had on any of my 6 Edis KVMs in the past several months was Iceland this week due to a DDoS attack on someone on the node but the downtime was minimal. By contrast the OVH Roubaix VPS (which will not be renewed) I'm testing rebooted twice during the past week. and the ChicagoVPS Atlanta VPS I canceled this week (and switched to RamNode which is much more stable) had rebooted twice in the week before I canceled it.

    I think you might be comparing apples to oranges, OVH is known to be the hotbed for all the skids and all that, and ChicagoVPS is just...cheap.

    On the other hand, EDIS (may be due to the fact that they are charging in Euros), sounds more like a premium provider from what I can tell, yes, I know they are still LEB and yes I know they have some good promotions, but in general I don't think they charge rock-bottom prices and instead tried to make sure they have a good reputation around.

    http://BornIn.Asia - FREE shared hosting and subdomain service for LET members! Click here to see how to get one yourself!
    96Forum: Low End VPS Discussions. Selling domains with GApp with various user counts (10 year reg incl. for some)

  • @William said: The only really stable bare metal virtualization seems to be VMWare, which runs stable under any amount of packets/mbit.

    That's why they are the best.

    Cheap VPS - VPSDime

  • OpenVZ is stable with .18 kernel.
    Our Xen/KVM never rebooted, nor locked up. OVZ did more times than I have fingers at both hands. True, they are also the most servers, but still, in total KVM+Xen are about half the number of OVZ and should have rebooted at least 5 times to keep the proportion.
    We do have DDoS too, but maybe not so much and is quickly nulled.
    On the other hand, yeah, IRC/Camfrog/Gameservers are here too the main problem.

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • DomainBopDomainBop Member
    edited April 2013

    OpenVZ is stable with .18 kernel.

    see below (from an IntoVPS openvz box)

    uptime

    20:19:49 up 179 days, 10:44, 0 users, load average: 0.07, 0.02, 0.00

    uname -a

    Linux xxx.xxx.xxx 2.6.32-308.el5.028stab099.3 #1 SMP Wed Mar 7 15:56:00 MSK 2012 i686 GNU/Linux

    .

    On the other hand, yeah, IRC/Camfrog/Gameservers are here too the main problem.

    I think that (the customers) is the main problem, not openvz/kvm/.18/.32 etc.

  • rm_rm_ Member
    edited April 2013

    @William said: 25% IRC servers

    So IRC servers, not even clients or bouncers, are only 25% of DDoS targets, yet you immediately single-out IRC as root of all evil (and readily get helpful suggestions to ban all IRC to solve everything).

    @jon617 said: Now I know why many providers disallow IRC.

    @Voss said: Perhaps it is time to re-consider that policy?

    IRC is the scapegoat, yeah, it's all because of IRC.

  • MaouniqueMaounique Member
    edited April 2013

    @DomainBop said: see below (from an IntoVPS openvz box)

    So, what, some nodes get lucky:

    [[email protected] ~]# uptime

    18:25:17 up 336 days, 4:17, 1 user, load average: 0.72, 0.83, 0.75
    [[email protected] ~]# uname -a
    Linux pm16.prometeus.net 2.6.32(...) Tue Apr 17 23:56:34 BST 2012 x86_64 x86_64 x86_64 GNU/Linux

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • I prefer to reboot now and then instead of run ancient kernels ;-)

    -

  • MaouniqueMaounique Member
    edited April 2013

    @rds100 said: I prefer to reboot now and then instead of run ancient kernels ;-)

    It is not that they are ancient, I am sure they are patched in time for vulnerabilities, however, some features are missing or need horrible (more so than mainstream OVZ, lol) patches to pretend to work.

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • jon617jon617 Member
    edited April 2013

    On the topic of reboots, I am fine with occasional reboots of virtual servers especially if it means upgrading to more stable kernels on the host machine. I would prefer, however, to be notified of such planned reboots in advance, even if the advance notice is just 15 minutes. That way at least I know why it was rebooted, without me having to dive into logs and open a support inquiry.

    EDIT: Just found http://status.edis.at. I will keep an eye on it for notices.

    On DOS attacks and KVM, this is good to know that KVM's network drivers don't react well when the host machine's network is overloaded.

    On why my virtual machines sometimes do not come back online after reboot, I am working with EDIS support right now on fixing this problem. FYI it seems to be an issue that likely would affect other customers who initially installed their OS by mounting an ISO image then left it mounted after the install was finished.

  • I've had a VPS at EDIS Swedish DC for about a year.
    My uptime record is 32 days. There are frequent reboots, atleast 2 per month. It also seems like they are getting DDoS a few times every week so you can't run anything that requires a reliable connection. I guess you get what you pay for!

  • rm_rm_ Member
    edited April 2013

    @MxMcMt said: I guess you get what you pay for!

    Oh right, "one LEB provider has a problem = any cheap VPS must also have it."
    I have news for you, nope, it's not like that.
    On my current Xen/KVM VPSes (not EDIS) I can easily have multiple months of uptime, if I wouldn't reboot them myself to install a kernel update. A sudden unannounced reboot from the provider is very much out of the ordinary for me, and if this even happens more than once, grounds to think about cancellation of that service.

  • @rm_ said: I have news for you, nope, it's not like that.

    Obviously ;) Some issues are out of their control and some aren't.. I'm not saying every cheap host has problems. Can't really complain about a few reboots when I only pay 3 EUR per month.

  • ChanChan Member

    I don't really mind if my servers reboots itself from time to time as everything restarts itself at boot anyway. I do mind when they deadlock and I have to force a hard reset from the control panel though, so far I've only experienced this once in my past year with EDIS but I think this should be improved.

  • TheLinuxBugTheLinuxBug Member
    edited April 2013

    First off, please do not take this as a bash towards Edis, but until recently their node in Spain was almost unusable. Constant reboots, network drop outs, sluggish system responses. However, part of that is my fault as its a yearly deal and I haven't found the time yet to bother with tickets, so maybe the node I am on is abused. It is still useful for what I needed it for at the moment, network traces and throughput tests (read: downloading test files 100mb in size from other places to test the speed to this location). I can not say I would ever put a site on that server or expect it to actually be able to serve anything reliably, it just hasn't been stable enough.

    My KVM in Switzerland however is rock solid and works without a hic-up, so I guess its just the luck of the draw and getting on a server without all the noisy neighbors.

    my 2 cents.

    Cheers.

    Have an Allwinner H3 device? Android? Check out H3Droid! | Lichee Pi Zero - The 6$ SBC | #SYSarm - Get It! | Atomic Pi - $35 x86 SBC
    21+ Years IT Experience in Linux/Windows Hosting, Administration and Development Services

  • WilliamWilliam Member, Provider
    edited April 2013

    @rm_ said: So IRC servers, not even clients or bouncers, are only 25% of DDoS targets

    No, it's mainly Bots and mainly Efnet it seems, not unlikely IRC spammers rather than real users however.
    And no, we don't plan to ban any IRC activity except servers (which are actually banned in most locations anyway), don't worry.

    I once watched #politics on Efnet also, seen someone say something idiotic - he ping timeouts, i see his host (149.*), seconds later Chicago went nearly entirely down, so there certainly are real users getting ddosd by others...

    EDIT:
    The shutdown KVMs after reboot of the node (no automatic start) should be fixed already on most nodes.

    @TheLinuxBug said: First off, please do not take this as a bash towards Edis, but until recently their node in Spain was almost unusable

    Yea, the ISP in Spain is... not very intelligent, but there is no replacement (trust me, we tried hard - even sending a spanish guy to call up local ISPs - with no luck), we now have more nodes and uplink capacity in Spain (and Poland, and France next week, and Switzerland) so these should be better now (along with less lockups as the vps are better spread out).

    Austria gets another 10G link and a new router end of April which will solve past DDoS issues.

    Sweden is... Sweden, network outages are mainly not caused by us but by other portlane customers getting attacked (Portlane hosts a lot of 'questionable' content, VPNs and IRC).
    Inevitable (this causes the now rather prevalent lag and ping spikes) for now, i guess Portlane will add additional capacity in the future.

  • ChanChan Member
    edited April 2013

    @William said: EDIT:

    The shutdown KVMs after reboot of the node (no automatic start) should be fixed already on most nodes.

    good to know thanks

    +1 for transparency

  • @William
    I have other stuff with Portlane directly, i have yet to experience any outages in the last 4 months or so. I call semi-bullshit.

  • Thanks @William for being honest and thorough once again.
    Edis may not be the best but never been disappointed me so far in more than a year, if not 2.

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • WilliamWilliam Member, Provider

    @blergh_ said: I have other stuff with Portlane directly, i have yet to experience any outages in the last 4 months or so. I call semi-bullshit.

    Likely depends on network segment and location (most Portlane customers are in Pionen, we are in Nacka which likely has less transport capacity to Pionen) and from what i would guess they run a single 10GE link per rack for shared colo.

  • @William do you allow IRC servers on other locations besides Austria?

    Hello, my name is Cloromorpho and i am a LEB addict.

  • WilliamWilliam Member, Provider

    No, but people still run them.

  • cloromorphocloromorpho Member
    edited April 2013

    @William If i run some little ircd leafs on other locations would you get mad and suspend my boxes?? xD

    Hello, my name is Cloromorpho and i am a LEB addict.

  • WilliamWilliam Member, Provider

    Yes, if noticed.

  • perennateperennate Member, Provider
    edited April 2013

    I haven't had any problems with the VPS exactly, but it does show huge bandwidth for some reason, even though the VPS is idling currently. Like 900 KB/s (summing incoming+outgoing and based on ifconfig), and no extra TCP connections. I tried checking the source with iftop (presumably UDP?), but it reports that root doesn't have permission to monitor :/

    Edit: contact form isn't working for me and I don't see ticket system

  • I don't see ticket system

    there isn't one. email support at edis.at

  • @perennate they do not run OVZ, but vserver, you see all node's stats.
    Unless you are on KVM, but from what you describe, i think not.

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • Htop on vservers, awesome view!

    Hello, my name is Cloromorpho and i am a LEB addict.

  • @William said: Yes, if noticed.

    Why only Austria?

    Hello, my name is Cloromorpho and i am a LEB addict.

  • Maybe pipes in other locations are not that "fat".

    Extremist conservative user, I wish to preserve human and civil rights, free speech, freedom of the press and worship, rule of law, democracy, peace and prosperity, social mobility, etc. Now you can draw your guns.

  • WilliamWilliam Member, Provider

    @cloromorpho said: Why only Austria?

    Because we have control here at attacks mainly.

  • perennateperennate Member, Provider
    edited April 2013

    @Maounique said: Unless you are on KVM, but from what you describe, i think not.

    I have no idea what I'm on, but I got reply from them and you're right, thanks. Guess that explains why iftop doesn't work. Although not sure how I'm supposed to make sure I'm using too much bandwidth then.

    I didn't know "vserver" (or I guess Linux-VServer?) was another virtualization software.

Sign In or Register to comment.