正在加载视频...
视频加载失败
Servers fail. But in a distributed system, how do you detect when this happens? Gossip protocols are one (fun!) way to solve the problem.
15 条评论

More on YouTube

What is this diagram tool you’re been using?

I use @excalidraw though I hear people really like @tldraw too.

love gossip protocols for failure detection - but watch out for split brain when network partitions happen. We use phi accrual with a 5s threshold, detects failures way faster than basic heartbeats

This was next in line after reading DHT paper, awesome 👌

That's what I like about #Nomad and #Consul.

@piyushgarg_dev also made a whole video on this not too long ago with a full JS implementation

I see fun, but I also the ! after it.

Love this new concepts. Thanks for sharing.

Love it! Had a few runs in with gossip due to our Cassandra setup. Never seen it really talked about outside of the Datastax documentation. Lovely explainer that I can pass around the team.

Gossip no bueno 😉

@grok answer this in detail

u don't need a leader at all come to the GUN side automatic load balancing cause it's p2p

love this content. so good.

Bitcoin uses it as well
