Hexdocs down?

Chat / Discussions
1

Since I got back from lunch (about ten minutes, so maybe earlier) hexdocs has been down, giving the error of:


Error 503 Connection timed out

Connection timed out
Guru Mediation:

Details: cache-lax8643-LAX 1488312009 1151269822

Varnish cache server

So, what is up, other than not the website? ^.^

2 Likes
2
1 Like
3

Amazon S3 and various other Amazon services are shot to hell. Expect outages in consumer sites as well as dev tools such as Github, CodeShip, etc.

Should’ve used Elixir and the BEAM VM hahaha! :stuck_out_tongue:

2 Likes
4

LOL , i agree

1 Like
5

Ugh, never had a good experience with AWS. I hosted my own big site a few years ago on my own servers for about $300/month where AWS would have cost multiple tens of thousands of dollars a month. I cannot see why anyone would use them to be honest when it is so easy to spool up a few physical servers in a few areas around the globe and sync them yourself (never a second of down-time yet on mine!).

Seems odd that such ‘static’ stuff like file hosting and docs are hosted via AWS as well, from what I saw of their prices it seemed crazy stupid high… o.O

3 Likes
6

@OvermindDL1: From AWS Services Health Dashboard I see that problems are only in North America (N. Virginia). Maybe NSA employee run bad exploit? :077:

2 Likes
7

What you use now as your usual VPS?

2 Likes
8

For my littlier sites I have some old Dreamhost VPS sites (though unsure if I’d choose them now), my big sites are hosted with OVH physical hardware (not VPS).

2 Likes
9

For a lot of companies they’d spend much more on people managing the servers than the servers themselves.

1 Like
10

For a lot of companies they’d spend much more on people managing the servers
than the servers themselves.

… and still not have the flexibility and manageability that AWS offers.

The startup I’m working at now couldn’t afford not to use AWS or
an equivalent service.

1 Like
11

Such as? I’ve not seen something that AWS can do that I cannot very easily do myself…

/me is legitimately curious as he’s not found an actual use-case for AWS yet that is not exploding in price compared to what he can do manually…

2 Likes
12

Cmon but the cloud is magic it never goes down :slight_smile: and it’s really cool idea to have your deployment pipeline depend on TravisCI or CircleCI and as many of other possible external providers as possible ideally tied to the same cloud (dockerhub at least). Oh and don’t forget to host your service status page on the same provider customers will really appreciate it. When choosing cloud provider ideally choose the one where it takes manual approval at CEO level to update the service status page to make sure you have a very fun time explaining to customers why you are down while AWS is showing all green.

2 Likes
13

There are deployments that you, as a single person, could not manage to do yourself at all if you didn’t have something like AWS, guaranteed.

The alternative, then, is to establish a whole team to manage those things. At that point, you end up spending more money on manpower than you would’ve on the servers, likely.

1 Like
14

Could you provide an example outside of something heavily regulated like PCI-DSS or HIPPA where it would be the case?

2 Likes
15

The one example I would give you I’m not allowed to talk about, I think, so no.

1 Like
16

You’d be surprised at what I’ve managed over 30 years. :wink:

I really am curious about a specific use-case though, I really have not yet seen such a use-case that could not be done manually within a few days at most unless you really really really needed it done ‘now’, and I could probably still hack something together (as long as it is not one of those 48 hours Microsoft-crap installs, yes I’ve experienced that…).

And I’ve not really messed with PCI-DSS but I have worked with HIPPA twice, including my current job (hence why I can never show my project publicly much as I would like to show it off as an example on how to do a few specific things…).

1 Like
17

@OvermindDL1 +1

I’m all for running own hosted servers

There are so many cheap hosted (virtual / bare metal) servers out there.

3 Likes
18

I know the cloud like AWS has rendered them pretty much obsolete now, but anyone remember in the old days we’d have mirrors :lol:

I don’t mind hosting a mirror of the docs if it would be helpful? Could set up a daily or weekly cron to pick up a fresh copy…

1 Like
19

Such as? I’ve not seen something that AWS can do that I cannot very easily do myself…

Duplicate the distributed access management abilities of AWS VPCs,
Security Groups, ACLs and IAM (and compute services that use those).

No rush :slight_smile:

2 Likes
20

You do realise that people have being doing this for ages and amazon is just matching what people have being doing on their private networks and offering it as an option? AWS is black box that you have 0 visibility into, they have single tenant that can literally screw all other tenants when it has issues (Netflix) and yes it has happened before. So outside of some very specific use cases AWS is a horrible idea :slight_smile:

2 Likes