Worker, raft and snapshot logic

Users
1

Hello,

I have a question about internal raft snapshots logic on the worker node.

I can see that:

  1. Save hard state and snapshot to WAL log (dgraph/draft.go at 146fb535f3d794f3f9ad25ac21fe640b875263f6 · dgraph-io/dgraph · GitHub)
  2. Then you try to sync using retrieveSnapshot function (dgraph/draft.go at 146fb535f3d794f3f9ad25ac21fe640b875263f6 · dgraph-io/dgraph · GitHub)

My question is what if node crashes during 2 step. So metadata saved, but database is out of sync with raft metadata.

2

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.