5 articles.
Leader election and leases answer a comforting question: who should be in charge right now. They do not fully answer the dangerous question: who can still write right now. A node can lose its lease…
Indefinite locks belong to a world where processes never crash and networks never split. That world does not exist. In a distributed system, “I hold the lock” can mean…
If your service runs on more than one node and still has a single instance assumption, you already have leader election. You just do not have it on purpose. Leader…
Distributed systems rarely fail because you picked the wrong cloud service. They fail because two nodes believe they are in charge, both act, and both are “correct” from…
Distributed systems do not fail because you missed a feature. They fail because responsibility is unclear. Two nodes act, both think they are right, and your data…
