13:49 - first errors
13:55 - first internal report
14:00 - status page update (acknowledgement)
14:05 - status page update (identified)
14:06 - customer support acknowledgement in chat
14:08 - last errors
14:16 - status page update (fixed & monitoring)
14:40 - status page update (resolved)
so, a ~20 minute outage is bad, but man is it a night-and-day difference level of communication and updates than certain other well known status pages.
Strange possible coincidence yesterday evening my iPhone screen kept going on by itself and it would display (with a play button) some weird URL that started with https://airtv and then had fastly somewhere in it.
Connected? I know, vague. It's just I've never heard of fastly until yesterday when this happened and today I see some kind of outage notice on hackernews :)
Considering the way people toss around the cloud services they use like they were merit badges in the HN commments, I'd say no, if it is working right, someone who frequents this site should have definitely heard of it..
Yeah I was disappointed that Imgix status page was 10+minutes behind Fastly, thankfully through Twitter/HN we were able to get a grasp of the situation before Imgix updated their all green page or responded to our tickets. It shocks me that a large provider like Imgix hasnt got a simple AWS Lambda timer, or Azure Function or whatever Google version is to check the status api or even scrap their upstream providers status page using say Cheerio and post something like 'we are investigating a potential problem, standby' orange alert till a human to get the bottom of the full picture. And yes, I consider 10+ minutes to publicly acknowledge to be poor for any large scale SaaS. I guess this attitude is related to a negative experience years ago with popular transaction email as a service provider that sounds like a poststamp that went down and didn't get fixed to their techs woke up many hours later (albeit all mail was queued - they were still 200ing err.. ) - I want to know you're on it and awake.
Their conference is today too: https://www.fastly.com/altitude/
They provide DNS and possibly more for Rubygems.org
Gonna be a bad day for a lot of folks if that goes down too.
They're also in front of PyPi and NPM.
Someone shared graphs of the drop in requests to PyPi at https://twitter.com/EWDurbin/status/880071713874202624
as a customer:
so, a ~20 minute outage is bad, but man is it a night-and-day difference level of communication and updates than certain other well known status pages.
Strange possible coincidence yesterday evening my iPhone screen kept going on by itself and it would display (with a play button) some weird URL that started with https://airtv and then had fastly somewhere in it.
Connected? I know, vague. It's just I've never heard of fastly until yesterday when this happened and today I see some kind of outage notice on hackernews :)
If it's working right, you wouldn't hear about it ;)
And if it's broken... well, yeah, I'd expect dependencies with poor error handling to behave unexpectedly.
Considering the way people toss around the cloud services they use like they were merit badges in the HN commments, I'd say no, if it is working right, someone who frequents this site should have definitely heard of it..
Includes services that use Fastly like Imgix.
Yeah I was disappointed that Imgix status page was 10+minutes behind Fastly, thankfully through Twitter/HN we were able to get a grasp of the situation before Imgix updated their all green page or responded to our tickets. It shocks me that a large provider like Imgix hasnt got a simple AWS Lambda timer, or Azure Function or whatever Google version is to check the status api or even scrap their upstream providers status page using say Cheerio and post something like 'we are investigating a potential problem, standby' orange alert till a human to get the bottom of the full picture. And yes, I consider 10+ minutes to publicly acknowledge to be poor for any large scale SaaS. I guess this attitude is related to a negative experience years ago with popular transaction email as a service provider that sounds like a poststamp that went down and didn't get fixed to their techs woke up many hours later (albeit all mail was queued - they were still 200ing err.. ) - I want to know you're on it and awake.