GitHub - Major Service Outage
status.github.comI would very much like to see a revival of this project:
https://code.google.com/p/gittorrent/
For a variety of reasons.
Seems to be one of the most severe Github service interruptions I've ever encountered. I can't even pull up the unauthenticated home page. I can't push, can't pull, can't do anything.
The D in github stands for decentralised.
But there is no D in github.
...oh.
just out of curiosity, does anyone find these HN postings useful?
wouldn't your critical shit break long before you go read about it on HN? otherwise, why not just rely on their human-friendly status pages that get served if such a thing occurs?
i don't understand why this is on the front page, since it's not an infrequent occurrence.
Misery loves company.
Also the posting comments become a 'place' to put suggestions, alternatives, and other measures to deal with the issue.
this is valid, though for stuff that occurs so frequently there should just be a sticky that can accumulate suggestions, not a clean slate submission for every time it happens.
There is a reliable indicator of whether or not people found the post useful - the number of points the submission has. To be honest, I find comments like yours far less useful, and they are not infrequent either.
i find this indicator not to be consistently reliable. anything that gets to the front page (which will never be pure crap), will always get upvotes.
but many (all?) submissions that make it to the front page get there by being manually curated there, not by accumulating upvotes first. on the flip side of the same coin, a lot of good things never make it to the front page, even if they gain a lot of points.
Major bummer when all of those open source projects go offline.. Does anybody do a mirror of github, either http or git-only?
Github's awesome, but yah, that's the downside of having one place that hosts so much code. Having some outage or disruption is feeling like a bi-weekly event..
"bi-weekly"... Do you mean twice a week or fortnightly?
I thought that ambiguity of meaning only applied to "bi-monthly", but a quick search shows I was wrong.
> "bi-weekly"... Do you mean twice a week or fortnightly?
On a previous team we developed the mental image of a bicycle, consisting of two cycles.
It seemed to help at the time.. but now I think about it I'm not sure how it did!
19:37 UTC - We are currently recovering from a database failure and have successfully failed over to a backup server.
A successful failure event probably should go unnoticed :)
I think someone should write a tutorial about how to continue working while Github goes down. If you're trying to deploy from a master branch on GitHub, this could be an issue, but if you're just trying to share projects between peers, this should be a that big of a deal right?
We found out the hard way that directly deploying from github might be a bad idea. I wrote a little web service that receives notifications via hooks when code is tagged and rehosts the tarballs in S3 for deployment. it's been rock solid in production.
Well this isn't good. Our integration server runs an npm install which includes a ton of dependencies that are github URLs. I guess tying our integration/deployment step to github isn't such a good idea after all.
"Somebody" needs to write a git caching proxy. Google searching comes up with a couple of attempts, neither looks production quality.
I'd suggest a rule of thumb where a build ought to be able to complete without any resources that you don't control the uptime of. As the number of such resources increases, the amount of time the build can not actually be built will tend towards 100%. A git caching proxy would be great for that sort of thing. A generic DVCS proxy would be even better.
No not at all. We keep local mirrors of everything that we pull in and integrate off that. Never a problem.
AAAAND we're back.
in other news, it must really suck to be github where your service being down is front page HN news.
The outage itself isn't great, but on the flip side you know you've made it when people give a shit about your downtime.
19:23 UTC. kken synchs the initial commit for his new repository.
ooops?
An hour after the 3D File announcement...cue unlikely/conspiratorial theories now.
Not a bad guess, honestly. Pretty sure some past outages have been due to new services (such as the new Search), either due to bugs or potential security issues.
I'm not trying to blame Github though, they're awesome!
Github was down so I took a break at work to read some Hacker News. "Ooh 3D file diffs, cool!" Click the link to read about the new feature, and dammit it's down too because the blog is hosted on github.com!
Its back now
I tried to git push; I kept getting 502 errors. I though, "Why isn't it working?" I came here to confirm it was down.