rust/neon
1
0
Fork 0
mirror of https://github.com/neondatabase/neon.git synced 2026-06-03 21:40:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

Reduce number of flushes in PostgresBackend by delaying flush till the lst prefetch request

Browse Source
This commit is contained in:
Konstantin Knizhnik
2022-10-22 15:38:09 +03:00
parent 60ef1d87f9
commit ab7a4d8574
3 changed files with 11 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
pageserver/src/page_service.rs
View File

@@ -300,7 +300,7 @@ impl PageServerHandler {
trace!("query: {copy_data_bytes:?}");
let neon_fe_msg = PagestreamFeMessage::parse(copy_data_bytes)?;
let mut do_flush = true;
let response = match neon_fe_msg {
PagestreamFeMessage::Exists(req) => {
let _timer = metrics.get_rel_exists.start_timer();
@@ -312,6 +312,7 @@ impl PageServerHandler {
}
PagestreamFeMessage::GetPage(req) => {
let _timer = metrics.get_page_at_lsn.start_timer();
do_flush = !req.prefetch;
self.handle_get_page_at_lsn_request(&timeline, &req).await
}
PagestreamFeMessage::DbSize(req) => {
@@ -330,7 +331,9 @@ impl PageServerHandler {
});
pgb.write_message(&BeMessage::CopyData(&response.serialize()))?;
pgb.flush().await?;
if do_flush {
pgb.flush().await?;
}
}
Ok(())
}

1
pgxn/neon/pagestore_client.h
View File

@@ -60,6 +60,7 @@ typedef struct
{
NeonMessageTag tag;
bool latest; /* if true, request latest page version */
bool prefetch; /* if true, then request is followed by more prefetch requests */
XLogRecPtr lsn; /* request page version @ this LSN */
} NeonRequest;

5
pgxn/neon/pagestore_smgr.c
View File

@@ -211,6 +211,7 @@ nm_pack_request(NeonRequest * msg)
NeonGetPageRequest *msg_req = (NeonGetPageRequest *) msg;
pq_sendbyte(&s, msg_req->req.latest);
pq_sendbyte(&s, msg_req->req.prefetch);
pq_sendint64(&s, msg_req->req.lsn);
pq_sendint32(&s, msg_req->rnode.spcNode);
pq_sendint32(&s, msg_req->rnode.dbNode);
@@ -383,6 +384,7 @@ nm_to_string(NeonMessage * msg)
appendStringInfo(&s, ", \"blkno\": %u", msg_req->blkno);
appendStringInfo(&s, ", \"lsn\": \"%X/%X\"", LSN_FORMAT_ARGS(msg_req->req.lsn));
appendStringInfo(&s, ", \"latest\": %d", msg_req->req.latest);
appendStringInfo(&s, ", \"prefetch\": %d", msg_req->req.prefetch);
appendStringInfoChar(&s, '}');
break;
}
@@ -1141,6 +1143,7 @@ neon_read_at_lsn(RelFileNode rnode, ForkNumber forkNum, BlockNumber blkno,
NeonGetPageRequest request = {
.req.tag = T_NeonGetPageRequest,
.req.latest = request_latest,
.req.prefetch = false,
.req.lsn = request_lsn,
.rnode = rnode,
.forknum = forkNum,
@@ -1150,6 +1153,7 @@ neon_read_at_lsn(RelFileNode rnode, ForkNumber forkNum, BlockNumber blkno,
if (n_prefetch_requests > 0)
{
/* Combine all prefetch requests with primary request */
request.req.prefetch = true;
page_server->send((NeonRequest *) & request);
for (i = 0; i < n_prefetch_requests; i++)
{
@@ -1157,6 +1161,7 @@ neon_read_at_lsn(RelFileNode rnode, ForkNumber forkNum, BlockNumber blkno,
request.forknum = prefetch_requests[i].forkNum;
request.blkno = prefetch_requests[i].blockNum;
prefetch_responses[i] = prefetch_requests[i];
request.req.prefetch = i+1 < n_prefetch_requests;
page_server->send((NeonRequest *) & request);
}
page_server->flush();