Hacker News Reader

Comment by maccard | Hacker News Reader

The common complaint is that when github goes down so do issue trackers/project management/docs etc. What are the equivalent alternatives for these systems that are distributed, so tolerant to downtime.

If you are in a situation that you're using a DVCS but relying on centralised everything else, and saying that you're unable to work because your Issue tracker/CI server is down, have you really gained anything over the alternatives?

View on HN

http://fossil-scm.org/

Thanks for the reasonable response - I've heard of fossil, but never had any experience with it.

You can setup gitlab with clones, if one server goes down, second has a copy. You know, git is a distributed system, unlike github.

But if two systems are identical, wouldn't it be likely that both would break at the same time?

No. Unless you do something dumb like "don't test that the failover system is fully functional before taking the primary offline;" this applies to any maintenance, which also includes upgrades. Usual HA tactics apply - don't rely on one power source, one network link, or just about any SPOF. (Which means that requirements start breeding like rabbits, which gets expensive rather quickly - whereas Github is free-as-in-beer, which is good enough for most people)

Perhaps the only case breaking both at once would be deliberate action (DDoS etc.).

Github is a distributed system, with a unifying front-end. The DDoS just ends up targeting all of it.

How do you propose to stop that with your proposal?

> You know, git is a distributed system, unlike github.

Thanks for the snark. I know git is a distributed system. My question was about everything else being down (your issue tracker, CI server, wiki/documentation), not about git itself.

Reply on news.ycombinator.com