"Unable to connect to application server." and phone and chat supports are down... What the flying f***?

#41

Anddd I’m down again.

1 Like
#42

There is no valid reason for it to be down. It would involve a lack of thinking on a cloud architect’s part.

Here’s how:

  • use a typical cloud supplier (they’re all similar, but I’ll use AWS as example)
  • pick a region near to the user, e.g. UK user pick London, US east coast pick Virginia, US west coast pick Oregon, etc)
  • in the region you use Availability Zones (AZs) and active-active load-balanced VMs (EC2 instances) for your web, application, database levels. You can use DBaaS like Redshift to ease the coding effort.
  • you replicate data in the background to another region (e.g. US west replicates to US East). If a region goes down you lose the last few hours of data, but in this Eufy case, the local storage in the camera / base stores last days worth anyway locally, so no actual data loss.
  • if a region were to go down, the Eufy application does not go down, you may see slightly out of date data.
  • DNS re-routing means there may be a slight hiccup of a few seconds, to a minute, at most.
  • if a VM goes down, the active-active failover kicks in, there is a brief loss of performance while another VM is spun up, takes 2 minutes a slightly slower service - autoscaling.
  • if you are doing software changes, use the “canary” method where a small % fraction of total users are exposed to the new version, if its unstable then send them back to the old version, run this over time til stability proved, then upgrade all users.

You’re welcome Eufy.

image
image.png922×697 128 KB

f3_0.png520×520 15.9 KB

3 Likes
#43

Agreed…though based on previous answers (or lack) I believe several parts of your points are not in place…even though it should be standard practice for web based applications…

#44

Does the developer’s mum know they’re working when should be at school?

(joking but it takes a seriously under-effort to bring a service down > minute in 2020)

1 Like
#45

True but look at some other aspects of issues mentioned on here…growing to fast with too small a team or…

#46

Have you guys tried emailing support?

#47

same issue…phone app says- Unable to connect to the application server.

2 Likes
#48

Mine is down too, what’s their customer service number? I need to call them :frowning:

2 Likes
#49

Perhaps it’s sleep time?

image
image.jpg1300×1211 138 KB

So there are very few reasons for this scale of failure is happening, the one which comes top of mind is they back-end to a non-cloud datacenter via a VPN and route all calls through the VPN. e.g. there’s a cable loose in Seattle.

1 Like
#50

Called support (800-988-7973) which had a pre-recorded message that said the call center was experiencing technical difficulties. Nothing more than that. Very disappointed.

1 Like
#51

That’s telling me there is a VPN and a spof (single point of failure) in the network architecture.

It is extremely difficult to have nil outages (involves syncronous replication which is expensive), but ones <1 minute is highly easy (async replication) to achieve. You just have to decide to. It involves redundant paths (cost) and timeout (configuration) to failover (coding). So costs.

IF the Eufy camera has local storage, it can spool locally and act locally while the network is down, if it copied from local to homebase, frequently, you can have the camera destroyed but all but last seconds in the base. The base can replicate to Cloud, and the internet can be done but you then just have local access.

It is all quite easy to code.

So this unreliability is a design choice, no accident.

1 Like
#52

I won’t be buying any more Anker/Eufy products. This is inexcusable for a security product.

1 Like
#53

All my recordings are on my base. I can see them, but they are slow to load into my phone app. I can see live video from the cameras. I cannot delete any old files because the application server is not working. I get “Failed to request (404)”

1 Like
#54

Wooah, don’t over react.

The Eufy security software and service failure is down to a small set of inexperienced folks. Don’t then avoid a whole parent company’s perfectly good folks. That is punishing yourself as a blunt weapon.

No, just don’t buy this particular product again, and don’t recommend it. Then you focus the avoidance to the under-engineering decision-makers solely and not generally.

#55

Once I was able to connect to the server I was able to install the 3rd camera. I also logged out with my phone and was not able to log back in while the server was having issues (could not see my camera feeds). I was still logged in with my tablet and was able to see the live feeds from my cameras but not adjust any settings until the server came back online. Things are working a little better at the moment now (I think the server is back up but may still be having issues. I don’t know).

1 Like
#56

And here I am again with a very expensive down system trying to figure out why.

3 Likes
#57

The engineer in me is doing

They wrote it to do it that way?

The canary software development method should mean one or two you may suffer an issue rarely but not repeatedly.

2 Likes
#58

You can do what you want, but this is not just limited to Eufy security products. I called the help number and received an automated message. This has been going on for hours and no update from the company. No explanation. This post was created over 4 hours ago. Is that acceptable for a security product to you?

1 Like
#59

Nope.

But I never bought it.

I mostly write my own software. I am familiar with the challenges.

The issue I’m more concerned about is not even today, it is the fact they don’t charge monthly fees. That is a timebomb. You get what you pay for.

Listen, venting doesn’t change anything, just make sure you don’t recommend it to anyone is all I’m saying. Software reliability and security is not that new, the constituent parts are all around the 10-15 year old. The trick to make it cheap and reliable is:

  • a) store the images locally in the camera until d)
  • b) copy the images to the homebase, this covers for a bad person destroying the camera on the outside
  • c) homebase copies images to local cloud region, this covers for the bad person destroying the camera, entering then destroying the base.
  • d) cloud region replicates to another region, then the camera can clear older images. This ensures images in at least 2 geographic areas and at least 4 datacenters. This covers for a bad day of the camera destroyed, base destroyed and a whole region down (say… a very large meteorological or seismic event as the cause).
  • e) allow the homebase to be accessed directly over LAN and WAN (dynamic IP) so if cloud down you have something.
  • f) allow the cloud copy to access to cover for home failure (it’s primary purpose)

I don’t see good engineering decisions here and I’m struggling to think why. Strikes me as inexperience. Just don’t buy the Eufy security products is the limit of the decision. Don’t paint all of Anker with the same brush, it’s totally different people engineering powerbanks, etc.

1 Like
#60

Sounds like you guys need to buy different cameras

2 Likes