Howdy, Stranger!

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


Digital Ocean Unreliable... Any replacement suggestion?
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.

Digital Ocean Unreliable... Any replacement suggestion?

Hello guys,

OK I just got sick of Digital Ocean 5 minutes ago... my VPS goes up and down from time to time and they just DON'T SEND ANY NOTIFICATION of an issue.

At the end, it cost me wayyy more money to have then behave like idiots (come on guys, the least you should do is send an email whenever a node is down so we can switch traffic to a rescue server) than moving elsewhere.

Now I'd like to hear your suggestions, my needs are pretty basic :

  • Full virtualization
  • VPS provider which DOES COMMUNICATE IMMEDIATELY when SHTF
  • Ideally with nodes in the US (+ South East Asia would be also good)
  • SSD nodes with some good disc size available (ideally in the 60-80GB range), around 4GB RAM / 2 CORES
  • Known players who've been around for a while

I got another VPS with Vultr which seems to be running OK but maybe I've been lucky. Others that come to mine are linode and ramnode, but not sure if any of these are better than the others.

Thanks a lot,

Cheers,

«1

Comments

  • It seems you want to replace automated monitoring with humans, who send you email when SHTF, so you can do whatever you do in this case, manually. Why? Why don't you just automate it all?

    Thanked by 3Saahib ATHK Pwner
  • BradBrad Member

    So am I correct when I assume that your budget is $40/mo?

  • IshaqIshaq Member

    I'm sure @jarland would be willing to help and take on suggestions :)

  • Every host down once in a while. You need monitoring for that. :o

  • @Brad not necessary $40, I would say in the $40-60 range, maybe even more if I got the feeling it's worth it.

    Yes I'm monitoring my App... but call me crazy but when a human does "on purpose" stop my VPS I expect him to send an email and say: "dude we have some issue, we'll stop your VPS now"... so far I would say it's that like 90% of my favorite providers do, so it doesn't seem to be the moon.

  • IshaqIshaq Member

    Go with a managed provider like WiredTree or KnownHost.

    Thanked by 1benjamin74
  • And if you want to laugh a bit, here is an excerpt of my support ticket:

    As for a notification, this is because our engineers normally send out an email of the event once the hypervisor issue has been resolved explaining the issue, as events like this normally last less than 4~ minutes but sadly this does not seem to be the case for your event.

    Ohhh, that's wonderful, so basically you send an email when it's solved... instead of sending one when it's broken...

    Once it's solved I can't do anything...
    If I knew when it's broken I could stop sending traffic to oblivion...

  • Awmusic12635Awmusic12635 Member, Host Rep
  • trewqtrewq Administrator, Patron Provider

    benjamin74 said: Ohhh, that's wonderful, so basically you send an email when it's solved... instead of sending one when it's broken...

    They say it normally takes 4 min to resolve, which would normally be a reasonable time frame for notification. Sometimes stuff goes wrong that is out of their control. What has your actual downtime been?

    Because this server is so important to your I'm assuming you can provide uptime and latency graphs.

    Once it's solved I can't do anything... If I knew when it's broken I could stop sending traffic to oblivion...

    Get your head out of your ass and either pay for that kind of service or setup your own fall over.

  • OMG, DO is unreliable!!!!! :D

    Thanked by 3ATHK yomero TheKiller
  • jarjar Patron Provider, Top Host, Veteran

    benjamin74 said: Ohhh, that's wonderful, so basically you send an email when it's solved... instead of sending one when it's broken...

    I agree, an email should be sent at the start of an issue and not at the end. Quite simply for that reason. Nodes can have issues, they're not perfect, and that's why you should take advantage of the features that we offer to reduce reliance on a single node:

    https://www.digitalocean.com/community/tutorials/how-to-create-a-high-availability-setup-with-heartbeat-and-floating-ips-on-ubuntu-14-04

    However, that you have to log in and check on things to know that there is an issue with the hypervisor (and not just your droplet) is a problem to me. We'll work on that. I'm sorry about the problem you experienced. Please know that we're not trying to "behave like idiots," we want your feedback on how we can improve and I think your feedback is identical to my own in this issue. If your hypervisor is truly going "up and down from time to time" and not just your web application (for whatever reason), something is not right and I would love to take a look at it.

  • bravo, spoken like a true ambassador :D

    @jarland said:

  • raindog308raindog308 Administrator, Veteran

    benjamin74 said: Yes I'm monitoring my App... but call me crazy but when a human does "on purpose" stop my VPS I expect him to send an email and say: "dude we have some issue, we'll stop your VPS now"...

    Hate to break it to you, but your expectations are unreasonable for a LEB provider.

    Yes, really.

    If you want that, get a managed provider.

    Ishaq said: Go with a managed provider like WiredTree or KnownHost.

    Exactly. Let's see, 4.5GB and 90GB of SSD is $85/mo at WiredTree, though you can find 10% coupons. I guarantee they'll notify you before any issues - and they answer the phone at 2am.

    @benjamin74 if you're expecting hand-holding service for $5/mo, just GTFO now.

    Thanked by 1Ympker
  • nepsneps Member

    My experiences with Digital Ocean have never been anything short that what I'd expect from the money I'm paying them. Looks like you need a managed service.

    benjamin74 said: (+ South East Asia would be also good)

    Also this. Almost all of the servers I buy are HK, SG or US West Coast, and once you cut out DO it's going to be hard to fill that Singapore hole. Some LE providers like HostUS maybe (who I also use, very happy with their service), but seriously, SG priced flat the same as everywhere else? A godsend.

  • SadySady Member

    If downtime costs you then go straight to AWS. It's simple.

  • jarjar Patron Provider, Top Host, Veteran
    edited April 2016

    @Sady said:
    If downtime costs you then go straight to AWS. It's simple.

    Even that's really only true if you're using all of their features like load balancers etc. It's important to note that there is no host (unmanaged) where you can count on 100% uptime without being willing to do the work to balance out your application across multiple servers.

  • @raindog308 said:
    if you're expecting hand-holding service for $5/mo, just GTFO now.

    You quoted him, but I don't think you read what he actually said.

    It is not unreasonable to expect to be notified on infrastructure-level customer-impacting events. He isn't looking for a host that will make sure his apache process is running and call him if it crashes, he just wants to be notified when there are infrastructure outages, planned or unplanned.

    Thanked by 1aglodek
  • Guessing many people are trying to burn their expiring DO credit? lol

  • TBH, just snapshot it, and then spin up another droplet with that snapshot. Probably is just your node having some hiccups.

  • CoreyCorey Member
    edited April 2016

    @Sady said:
    If downtime costs you then go straight to AWS. It's simple.

    Yes - AWS, where you get charged $10 per 100GB of traffic.

  • KIL86KIL86 Member

    raindog308 said: if you're expecting hand-holding service for $5/mo, just GTFO now

    You should probably re-read his posts. He said he just wanted a notification when a issue is found, not resolved. Also, he said that his budget was $40-$60 per month and that he would pay more for exceptional service.

    @benjamin74 as the others have said, a managed service provider is what you are looking for.

    Thanked by 1iKeyZ
  • jarjar Patron Provider, Top Host, Veteran

    @yomero said:
    TBH, just snapshot it, and then spin up another droplet with that snapshot. Probably is just your node having some hiccups.

    On the other side, it's one click + confirm for me to migrate it, and if a single HV is repeatedly failing and people haven't been migrated off of it I must say my interest is high.

    Thanked by 3yomero netomx aglodek
  • benjamin74 said: my VPS goes up and down from time to time

    I find this hard to believe. I've been with DO for over 2 years now and I've had (planned) down time literally once due to maintenance at the DC. How do yo know for certain that it's DO and not something in between?

  • ATHKATHK Member

    @grimsdottir said:
    I find this hard to believe. I've been with DO for over 2 years now and I've had (planned) down time literally once due to maintenance at the DC. How do yo know for certain that it's DO and not something in between?

    I've been with them a long long time, I've had multiple node reboots (notification after the fact) few planned network maintenances a few emergency fixes and planned reboots, never had downtime that I haven't been aware of, apart from the node reboots which don't last long.

    It's a solid service, though it would be great to get a heads up before rebooting the node, Jarland said he's looking into it so I'm sure we'll see that change in the future.

  • raindog308raindog308 Administrator, Veteran

    HackedServer said: You quoted him, but I don't think you read what he actually said.

    Actually I did.

    HackedServer said: It is not unreasonable to expect to be notified on infrastructure-level customer-impacting events. He isn't looking for a host that will make sure his apache process is running and call him if it crashes, he just wants to be notified when there are infrastructure outages, planned or unplanned.

    I don't think you read what he actually said :-)

    "my VPS goes up and down from time to time and they just DON'T SEND ANY NOTIFICATION of an issue."

    Right. And that's not surprising. First, they may not notice - what is "my VPS goes up and down". It could be a network issue, a DC issue, the user's home broadband, etc. He's looking for monitoring and that isn't really a typical LEB feature.

    "come on guys, the least you should do is send an email whenever a node is down so we can switch traffic to a rescue server"

    And that's unreasonable for $5/mo.

    I realize we're talking about DO and they're not truly a "LEB provider" in the same sense as many other hosts here - for example, they have more than 1-2 employees - but once you expect <24 hours response time for tickets/problems I think you're getting into unreasonable territory.

    Certainly there are LEB hosts that deliver a better experience...but we're talking about reasonable expectations.

  • iKeyZiKeyZ Veteran
    edited April 2016

    @raindog308 said:
    And that's unreasonable for $5/mo.

    Again, who is talking about $5 per month? It's not unreasonable to get notifications before (if known) or during downtime instead of waiting until after with his budget (which was $40-$60 per month).

  • shovenoseshovenose Member, Host Rep

    Alright, what if they added a Monitoring page to their panel. By default it could simply ping each VM every 5 minutes, and a user could enable like HTTP monitoring as well. This would be pretty easy to code I think and could be displayed easily with green check boxes and yellow warning signs so even stupid customers like the OP who think DigitalOcean is unreliable (LOL!) will understand it.

    And even WiredTree has downtime. They screwed up my server and it was down for an entire night. Though that made for a rather sleepless night, they gave me the whole month for free, so kudos to them. Highly recommended!

  • @Jarland:

    Had some speed issues with deploying a droplet from a snapshot, a 55-second deployment turned into a 2-hour panic-attack deployment, $160/mth plan...

    Opened a support ticket, was told that it took quite some time for the snapshot to copy over from the NAS.

    Maybe you wanna get your team to starting scaling the snapshot storage further?

    Here's a question though: How do I make sure that my nodes are deployed in different hypervisors, for a higher redundancy setup in the event of the hypervisor itself crashing?

    Otherwise, OP, DO is awesome, maybe this is just an one-off incident.

    Thanked by 1jar
  • jarjar Patron Provider, Top Host, Veteran

    @theroyalstudent said:
    Jarland:

    Had some speed issues with deploying a droplet from a snapshot, a 55-second deployment turned into a 2-hour panic-attack deployment, $160/mth plan...

    Opened a support ticket, was told that it took quite some time for the snapshot to copy over from the NAS.

    Maybe you wanna get your team to starting scaling the snapshot storage further?

    Here's a question though: How do I make sure that my nodes are deployed in different hypervisors, for a higher redundancy setup in the event of the hypervisor itself crashing?

    Otherwise, OP, DO is awesome, maybe this is just an one-off incident.

    Understandable and appreciate the feedback. Certainly a 55 second deployment isn't always possible with a snapshot given that snapshots can contain as much as 640GB of data. You should always make time for time sensitive deployments based on variables like that, just like I wouldn't start an rsync of 500GB of data knowing I needed to deploy something from it in 5 minutes.

    That said, it's still good feedback and I'm sure improvements could be made.

  • @raindog308 said:
    It could be a network issue, a DC issue, the user's home broadband, etc

    In his posts he has a statement from DO support that it was a hypervisor issue, so I don't think it was his internet.

    @raindog308 said:
    He's looking for monitoring and that isn't really a typical LEB feature.

    No, he already said he monitors his app. I imagine that's how he is aware that it goes up and down.

    @raindog308 said:
    but once you expect <24 hours response time for tickets/problems I think you're getting into unreasonable territory.

    I think there you just have very low expectations. If an entire DC went offline you'd be fine not being notified or updated to what was going on? And that's only part of what he is wanting, he wants to be proactively notified when the provider is going to purposefully impact his service. Unless it is accidental there should be no reason why clients aren't alerted that their service is being interrupted.

    And just because:

    What is your SLA? DigitalOcean provides a 99.99% Uptime SLA.

Sign In or Register to comment.