Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- Probably no synchro
- Focus on project 3, project 2, everything else
- Raft questions
- Maybe something on project 2- stack, virtual memory, syscalls
- 3 servers with correct implementations
- S1 is leader in term 3
- S2 and S3 are followers in term 3
- All logs have identical entries
- Servers can append new client actions (not like P3)
- Describe in 10 events or fewer a situation in which a server is in candidate mode receives a non-entry appendEntries RPC from a server, S', whose term is greater than or equal to S's, i.e., term of S' as leader >= term of S as candidate
- Events include: new updates, mode changes, failures/recovery, slow/failing network, paused server, machine can get slow/fail
- 1. S1 crashes (term 3)
- (S2 goes into candidate mode)
- 2. S2 elected leader (term 4)<---increment term on election
- 3. S2 receives new update (term 4)
- 4. S1 recovers w/o network (term 3)-cant be updated
- 5. S1 becomes candidate (term 4)- bc crashed servers always come back as followers
- 6. S1 network heals (term 4)
- 7. S2 forwards update to S1 (term 4) appendEntries
- When server is paused it cannot receive messages/appendEntries requests
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement