Brutal.

[ingoschi@digitalcourage.social]

@ironicbadger searching for "github downtime 2016" I don't trust the old values

[simonwaight@mastodon.online]

@ironicbadger interested if you can map that to growth/shrinkage in users and/or volume or data being used.

[luceos@fosstodon.org]

@ironicbadger wait, they went from 100% uptime on their own servers,which were constantly on, to 99,5% as lowest uptime on a scalable infrastructure? That's 3,5 hours max spread over a month. And you're taking the piss on them for that? That's wild. Maybe focus on their link to Israel in the Gaza war or something.

[hikhvar@norden.social]

@Jourei AFAIK they started to use Azure more and more. That would explain the initial stability after the buy event. Also they might have shifted from stability to ship feature after feature. Before Microsoft bought them, their feature set was rather stable.

@ironicbadger

[miyuru@ipv6.social]

@Tubsta Yep still 0% uptime from my IPv6 only server point of view.

@ironicbadger

[muan@mastodon.social]

@Jourei @ironicbadger No, when MSFT came in they tried to change everything at once. If anything services fee can be paid back into MSFT pipelines instead of other service providers, they wanted it yesterday. They brought in more than 1000 people, which was more than all of GitHub at the time, to dilute the original teams.

[follpvosten@karp.lol]

@ironicbadger lmao yeah that checks out

[zetasyanthis@dragon.style]

@ironicbadger

And this is why we keep local repo copies for production repos.

[luc0x61@mastodon.gamedev.place]

@ironicbadger Microsoft has always had a peculiar reverse Mida's touch, transforming into junk everything they touch - remember Zune, Windows Phone, etc.

[ironicbadger@techhub.social]

@thunfisch @isotopp https://techhub.social/@ironicbadger/116327363904452057

[crazyeddie@mastodon.social]

@Jourei @ironicbadger "restructuring"

They fired huge chunks of the existing staff and terrified the rest.

[mbpaz@mas.to]

@luceos @ironicbadger Critical services tend to measure availability (informally) in "nines". 99.999% is good enough for most purposes; run your critical system at 99.9% availability and your contract is immediately over.

3.5h downtime in a month is ridiculous. It's teenager-computer-in-a-closet availability.

[coldclimate@hachyderm.io]

@ironicbadger is that because people left, or tech investments stopped, or something like bits of housing were moved to azure?

[murb@todon.nl]

@ironicbadger @a_different_jlh when did they really start measuring though, the source of the graph also shows data before the invention of even git itself: https://www.githubstatus.com/uptime?page=200

[thunfisch@chaos.social]

@ironicbadger @isotopp Thanks! So it is indeed the data from the official status page. Yes, since microslop took over reliability has gotten even worse, but it was bad before as well. They just often did not mark the services as down on the status page, even though they were.

So, the general conclusion here is still correct, but the data for the visualization is not reliable IMO.

[aerique@genart.social]

@jmelesky @ironicbadger That was fun.

[comsey@mstdn.social]

@ironicbadger I expect them to start measure the SLA properly after the acquisition - such a change looks rather than a change of methodology rather than immediate effect of a merge (fuck Microsoft anyway, but not for that)

[skjeggtroll@mastodon.online]

@ironicbadger

No system with million of users has 100% uptime across two years.

I suspect this is more about what and how up-time is being measured.

[goleztrol@mastodon.social]

@luc0x61 @ironicbadger
Minecraft

[rberthier@oc.todon.fr]

@ironicbadger Hi, source please. Not only it's 1st of April, but I want to click on this "Breakdown" tab