Patrick Insinger 09e29c3e54 pageserver - fsync layerfiles after writing ...

Instead of fsyncing after each file is written, do all the fsyncs at the
end. On macOS I observed `zenith init` time go from 36 s -> 3 s.

2021-10-28 13:28:03 -07:00

..

bin

Fix toml serde gotchas

2021-10-18 14:14:27 +03:00

http

review cleanups

2021-10-27 13:30:34 +03:00

layered_repository

pageserver - fsync layerfiles after writing

2021-10-28 13:28:03 -07:00

relish_storage

Add core relish backup and restore functionality

2021-10-22 22:22:38 +03:00

basebackup.rs

Support read-only nodes

2021-10-19 09:48:12 +03:00

branches.rs

Refactoring of checkpointer and GC.

2021-10-27 20:50:26 +03:00

layered_repository.rs

pageserver - fsync layerfiles after writing

2021-10-28 13:28:03 -07:00

lib.rs

Refactoring of checkpointer and GC.

2021-10-27 20:50:26 +03:00

page_service.rs

review cleanups

2021-10-27 13:30:34 +03:00

relish_storage.rs

Add core relish backup and restore functionality

2021-10-22 22:22:38 +03:00

relish.rs

fix clippy warnings

2021-09-02 18:54:44 +03:00

repository.rs

Refactoring of checkpointer and GC.

2021-10-27 20:50:26 +03:00

restore_local_repo.rs

On new repo, start writing WAL only after the initial checkpoint record.

2021-10-19 09:48:04 +03:00

tenant_mgr.rs

Refactoring of checkpointer and GC.

2021-10-27 20:50:26 +03:00

tenant_threads.rs

Refactoring of checkpointer and GC.

2021-10-27 20:50:26 +03:00

waldecoder.rs

Optimize WAL decoding slightly.

2021-10-14 14:21:23 +03:00

walreceiver.rs

Implement backpressure for compute node to avoid WAL overflow

2021-10-21 18:15:50 +03:00

walredo.rs

Don't wait for wal-redo process for non-relational records replay

2021-10-26 19:30:28 +03:00