Hacker News new | ask | show | jobs
by lukepothier 1655 days ago
I thought I and all my colleagues were getting 500s, but then I checked the status page and saw a green check. Silly me!
4 comments

A status page properly done is just a link to HN.
That very much is what https://downdetector.com/ and https://downforeveryoneorjustme.com/ are, crowd-sourced downtime reports. HN as a whole doesn't care if some niche industry site is up/down, so would never surface past the 'new' page. It's also helpful to see if a problem is localized to a particular region.
Hah, I just did the same thing - before clicking through to the status page, I had a bet with myself that everything would be green. Imagine my surprise when indeed everything was green.

If companies like Amazon and Google are going to outright lie on their status pages, why bother having them at all?

Because somebody (customer) is monitoring the status page and measuring how long it was not green and hold you accountable for your service being down. So it‘s „better“ to update your status page afterwards when your legal team is all on board.
That doesn't answer the question of why bothering to have them at all.
Partially. Don‘t have one: customers complain. Have one, but with lies: customers complain. Have one that shows the truth: legal department complains.
I hope you appreciate that that breakdown also doesn't answer the question of why bothering to have them at all? ;P
I know youre joking, but he did, because "customers complain, but legal department doesnt"
Because it's very profitable to be able to say that you have 5 nines (99.999%) uptime, over your competition with 4 nines or less.
For the afterwards part, I assume

>So it‘s „better“ to update your status page afterwards when your legal team is all on board

Because somebody (customer) wants to monitor a status page
But that purpose was explicitly undermined by the status page not providing the actual status!
>„better“

Is the initial double quote at the bottom something that is language specific?

I did not explicitly create them. HN-app on iOS did them by itself.

And it‘s probably specifiy to German.

Technically, German should have them, but barely anyone bothers since they're not on a normal keyboard layout. Some word processors replace them automatically though.

Also, the upper one is “, not ".

Status pages of big companies like Google, Facebook etc. seems to never work. They are all green even when no one can access the service
Sounds like they don't have a status page, but just a placebo.
It probably causes no end of headaches fir IT departments and ISPs:

*"The site is up, I just can't get to it. Fix my network!"

I don't think status pages are so easy in practice. Me and everyone I know have not had any problems with google calendar today, so who is right? Together we're probably <0.001% of all google calendar users so I don't think our sample is very representative.

Presumably there is some internal threshold for when to flick the switch on the service dashboard, but without knowing anything about the scale of the outage or what the threshold is we're kinda shooting in the dark.

That's why you see the orange color instead of red. To indicate that 1. The survive is not fully functional, or 2. The service is not functional for some regions. And I've seen some companies add comments to each status to explain what's going on.
I'm sure the statuses are being very liberal at their reporting thresholds towards the end user. /s

Perhaps it would help if the dashboards gave a tiny inkling of transparency, like what the thresholds are so you could gauge the relative significance of your personal service outage. If my page is out and vendor shows green, then maybe those relying on my service ontop of vendor won't believe me as much when I pass blame to them and/or maybe the vendor isn't even aware of the issue, so I should contact them because I'm an edge case.

If my service is out and the status page shows red with a nifty "were working on it" I can very clearly show anyone I report to or who is dependent on my service the parent service outage causing the issue, maybe even link them to it. In addition, I know that I don't need to contact the vendors support because they're aware of the issue and actively working on resolving it. I'm sure their support staff will appreciate it that millions of people can avoid contacting their support service reporting the same outage and having to repeat the same information through more costly reporting mechanisms than a page everyone can simply observe.

Every power company I've been with for at least the past 10 years basically do this. If I'm out at work and come home to no power, I first load my power company's outage report site (service status dashboard). I check my area and magically I know if the power company is aware and likely working on it, I can even see their updated time estimates when service will be restored. Sometimes they even tell me what caused the issue (oh boy, information). I never even have to make a call.

Meanwhile if I'm sitting here sipping coffee and hear a loud noise down the street, power is out in the middle of the day and I look across the street and see the neighbor's interior lights on, check the power company's outage map and don't see it: I give them a call. Tada, I know my isolated outage will be addressed and the service provider is aware of the issue. I also know if it looks limited, I'm probably going to be a lower priority if there are multiple outages and limited maintenance staff, so I can sort of expect it may be several hours. I can head out and not waste my time and effort sitting around.

Let's treat critical internet service infrastructure (that's what 'cloud' wanted to be, so let them have all that entails) like we treat other critical service infrastructure. It works quite well.

The underlying issue I see is that my utilities are well regulated and watched over so there often is a lot of transparency about what's going on. Heck, they even have public hearings if they want to increase their prices. Meanwhile, private company offering some service isn't very regulated and has every incentive not to let you know to manage the perception of their image and reliability vs providing empirical actual data. So will we see any sort of transparency? I doubt it, just as I highly doubt the size of represntative truthfulness of any data they report.