Do not transfer WAL to computation nodes: use pg_resetwal for node startup

postgres_ffi/src/lib.rs

@@ -3,6 +3,7 @@
 #![allow(non_snake_case)]
 include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
 
+pub mod nonrelfile_utils;
 pub mod pg_constants;
 pub mod relfile_utils;
 pub mod xlog_utils;
@@ -11,6 +12,7 @@ use bytes::{Buf, Bytes, BytesMut};
 
 // sizeof(ControlFileData)
 const SIZEOF_CONTROLDATA: usize = std::mem::size_of::<ControlFileData>();
+const SIZEOF_CHECKPOINT: usize = std::mem::size_of::<CheckPoint>();
 const OFFSETOF_CRC: usize = PG_CONTROLFILEDATA_OFFSETOF_CRC as usize;
 
 impl ControlFileData {
@@ -69,3 +71,42 @@ pub fn encode_pg_control(controlfile: ControlFileData) -> Bytes {
 
     buf.into()
 }
+
+pub fn encode_checkpoint(checkpoint: CheckPoint) -> Bytes {
+    let b: [u8; SIZEOF_CHECKPOINT];
+    b = unsafe { std::mem::transmute::<CheckPoint, [u8; SIZEOF_CHECKPOINT]>(checkpoint) };
+    return Bytes::copy_from_slice(&b[..]);
+}
+
+pub fn decode_checkpoint(mut buf: Bytes) -> Result<CheckPoint, anyhow::Error> {
+    let mut b = [0u8; SIZEOF_CHECKPOINT];
+    buf.copy_to_slice(&mut b);
+    let checkpoint: CheckPoint;
+    checkpoint = unsafe { std::mem::transmute::<[u8; SIZEOF_CHECKPOINT], CheckPoint>(b) };
+    Ok(checkpoint)
+}
+
+impl CheckPoint {
+    pub fn new(lsn: u64, timeline: u32) -> CheckPoint {
+        CheckPoint {
+            redo: lsn,
+            ThisTimeLineID: timeline,
+            PrevTimeLineID: timeline,
+            fullPageWrites: true, // TODO: get actual value of full_page_writes
+            nextXid: FullTransactionId {
+                value: pg_constants::FIRST_NORMAL_TRANSACTION_ID as u64,
+            }, // TODO: handle epoch?
+            nextOid: pg_constants::FIRST_BOOTSTRAP_OBJECT_ID,
+            nextMulti: 1,
+            nextMultiOffset: 0,
+            oldestXid: pg_constants::FIRST_NORMAL_TRANSACTION_ID,
+            oldestXidDB: 0,
+            oldestMulti: 1,
+            oldestMultiDB: 0,
+            time: 0,
+            oldestCommitTsXid: 0,
+            newestCommitTsXid: 0,
+            oldestActiveXid: pg_constants::INVALID_TRANSACTION_ID,
+        }
+    }
+}

postgres_ffi/src/pg_constants.rs

@@ -23,15 +23,21 @@ pub const PG_XACT_FORKNUM: u8 = 44;
 pub const PG_MXACT_OFFSETS_FORKNUM: u8 = 45;
 pub const PG_MXACT_MEMBERS_FORKNUM: u8 = 46;
 pub const PG_TWOPHASE_FORKNUM: u8 = 47;
+pub const PG_CHECKPOINT_FORKNUM: u8 = 48;
 
 // From storage_xlog.h
 pub const SMGR_TRUNCATE_HEAP: u32 = 0x0001;
 
+// from pg_config.h. These can be changed with configure options --with-blocksize=BLOCKSIZE and
+// --with-segsize=SEGSIZE, but assume the defaults for now.
+pub const BLCKSZ: u16 = 8192;
+pub const RELSEG_SIZE: u32 = 1024 * 1024 * 1024 / (BLCKSZ as u32);
+
 //
 // constants from clog.h
 //
 pub const CLOG_XACTS_PER_BYTE: u32 = 4;
-pub const CLOG_XACTS_PER_PAGE: u32 = 8192 * CLOG_XACTS_PER_BYTE;
+pub const CLOG_XACTS_PER_PAGE: u32 = BLCKSZ as u32 * CLOG_XACTS_PER_BYTE;
 pub const CLOG_BITS_PER_XACT: u8 = 2;
 pub const CLOG_XACT_BITMASK: u8 = (1 << CLOG_BITS_PER_XACT) - 1;
 
@@ -42,6 +48,7 @@ pub const SIZE_OF_PAGE_HEADER: u16 = 24;
 pub const BITS_PER_HEAPBLOCK: u16 = 2;
 pub const HEAPBLOCKS_PER_PAGE: u16 = (BLCKSZ - SIZE_OF_PAGE_HEADER) * 8 / BITS_PER_HEAPBLOCK;
 
+pub const TRANSACTION_STATUS_IN_PROGRESS: u8 = 0x00;
 pub const TRANSACTION_STATUS_COMMITTED: u8 = 0x01;
 pub const TRANSACTION_STATUS_ABORTED: u8 = 0x02;
 pub const TRANSACTION_STATUS_SUB_COMMITTED: u8 = 0x03;
@@ -136,11 +143,6 @@ pub const XLOG_TBLSPC_DROP: u8 = 0x10;
 
 pub const SIZEOF_XLOGRECORD: u32 = 24;
 
-// from pg_config.h. These can be changed with configure options --with-blocksize=BLOCKSIZE and
-// --with-segsize=SEGSIZE, but assume the defaults for now.
-pub const BLCKSZ: u16 = 8192;
-pub const RELSEG_SIZE: u32 = 1024 * 1024 * 1024 / (BLCKSZ as u32);
-
 //
 // from xlogrecord.h
 //
@@ -162,3 +164,12 @@ pub const BKPBLOCK_SAME_REL: u8 = 0x80; /* RelFileNode omitted, same as previous
 pub const BKPIMAGE_HAS_HOLE: u8 = 0x01; /* page image has "hole" */
 pub const BKPIMAGE_IS_COMPRESSED: u8 = 0x02; /* page image is compressed */
 pub const BKPIMAGE_APPLY: u8 = 0x04; /* page image should be restored during replay */
+
+/* From transam.h */
+pub const FIRST_NORMAL_TRANSACTION_ID: u32 = 3;
+pub const INVALID_TRANSACTION_ID: u32 = 0;
+pub const FIRST_BOOTSTRAP_OBJECT_ID: u32 = 12000;
+pub const FIRST_NORMAL_OBJECT_ID: u32 = 16384;
+
+/* FIXME: pageserver should request wal_seg_size from compute node */
+pub const WAL_SEGMENT_SIZE: usize = 16 * 1024 * 1024;