Miscellaneous follow-ups for distributed Restate

[tillrohrmann]

Umbrella issue for collection various improvements related for the distributed Restate work.

Tasks

Preview Give feedback

1. Don't panic in case of failures but rather retry and report health status #1266
   2 of 2
   +0
2. Avoid blocking in the PartitionProcessorManager when being in the control loop #2146
   +0
   
3. Split ingress role from worker role #1312
   +0
4. Split runtime logging from service logging #1310
   onboarding +1
   
5. Metadata store requirement for bootstrap node not needed with external stores
6. Make TCP connection settings configurable #2153
   +0
   
7. Enable Restate runtime to trace across processes #2155
   +0
   
8. Investigate runtime behavior when TCP connections are dropped instead of being rejected #2156
   +0
9. Decouple ingress from PartitionStoreManager #1811
   +0
   
10. Harden cluster controller to not panic on error #2276
    +0
11. Replace infinite retries with writing to background appender when announcing leadership #2333
    +0
    
12. Slow append performance if nodes are down #2377
    +0
13. Prevent usage of Node Id 0 #2550
    +0
14. Split MetadataStoreOptions into LocalMetadataStoreOptions and MetadataServerOptions #2552
    +0
15. A more efficient serialization format for raft metadata store #2544
    +0
16. Include remote metadata-store address in error message #2554
    +0
17. Metadata store observability and tooling #2557
    +0
    
18. Improve cluster observability #2559
    0 of 5
    +0
19. restatectl cluster provision only respect --log-replication iff --provider is provided #2561
    +0
    
20. Setting the LogServer state from Provisioning to ReadWrite is not idempotent #2563
    bug +1
21. Restatectl cluster provisions requires me to specify log-default-nodeset-size #2604
    +0
    
22. Print nodes as provisioning when running restatectl status or nodes ls command #2609
    +0
23. Don't log details and metadata when logging Status error #2607
    +0