Every single post along these lines gets some comment about how it's acceptable/understandable because they've had such a massive influx of usage but I'm sorry at this point it's completely unacceptable for a service which acts as such a vital part of so many organization's workflows to have this level of downtime.
Given I've never worked at an organization at such scale, but there have been several other companies who have experienced large influxes of usage and been able to scale WAY more gracefully than github. It's not even like these outages are some delayed worker assignment due to higher supply than demand, it's just complete outages that make deployment impossible.
Simply unacceptable, hopefully this one doesn't take the NN hours it did last time to resolve.
Yeah I do not understand this line of thinking at all. I do not care if they’re inundated with exponential traffic because of LLMs.
If I was a betting person I would say that 99% of those new repos are not paying and yet they’re causing failures at an unheard of frequency. It seems so odd to me that they’re not rate limiting hard for non-paying users.
I would totally accept that as "okay" - but then, we pay for that service usage based, and it doesn't go both ways: its down, its down, nobody credits us for the hours it was down. Instead we are waiting, and looking at a status page, hoping it will resolve.
I am wondering if it would be a lot more stable if users would be credited every time it was down on a "usage basis"...
Every single post along these lines gets some comment about how it's acceptable/understandable because they've had such a massive influx of usage but I'm sorry at this point it's completely unacceptable for a service which acts as such a vital part of so many organization's workflows to have this level of downtime.
Given I've never worked at an organization at such scale, but there have been several other companies who have experienced large influxes of usage and been able to scale WAY more gracefully than github. It's not even like these outages are some delayed worker assignment due to higher supply than demand, it's just complete outages that make deployment impossible.
Simply unacceptable, hopefully this one doesn't take the NN hours it did last time to resolve.
Yeah I do not understand this line of thinking at all. I do not care if they’re inundated with exponential traffic because of LLMs.
If I was a betting person I would say that 99% of those new repos are not paying and yet they’re causing failures at an unheard of frequency. It seems so odd to me that they’re not rate limiting hard for non-paying users.
I feel for their support and infra teams when this happens.
Send them a virtual hug or nice supportive message, people, it'll brighten their stressful day.
Remember they are humans, like I _hope_ you are.
I would totally accept that as "okay" - but then, we pay for that service usage based, and it doesn't go both ways: its down, its down, nobody credits us for the hours it was down. Instead we are waiting, and looking at a status page, hoping it will resolve.
I am wondering if it would be a lot more stable if users would be credited every time it was down on a "usage basis"...
GitHub is down.
Time to pretend we were productive without it.
Some more discussion:
Disruption with Some GitHub Services
https://news.ycombinator.com/item?id=49450722
Ah shit, here we go again
are down ... oh wait