JakeHillion/storj
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

satellite/satellitedb: optimize ProjectAccounting.ArchiveRollupsBefore

Browse Source 
The previous query was making a full table scan. This modifies code to
do the queries separately on each action. It will probably be slower on
a small table, however there should be a several magnitude boost on
large tables.

Change-Id: Ib8885024d8a5a0102bbab4ce09bd6af9047930c9
This commit is contained in:
Egon Elbre 2021-07-02 16:41:49 +03:00
parent c248651f3f
commit bf5194d134
1 changed files with 40 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

72
satellite/satellitedb/projectaccounting.go
View File

@ -663,7 +663,7 @@ func (db *ProjectAccounting) GetBucketTotals(ctx context.Context, projectID uuid
}
// ArchiveRollupsBefore archives rollups older than a given time.
func (db *ProjectAccounting) ArchiveRollupsBefore(ctx context.Context, before time.Time, batchSize int) (bucketRollupsDeleted int, err error) {
func (db *ProjectAccounting) ArchiveRollupsBefore(ctx context.Context, before time.Time, batchSize int) (archivedCount int, err error) {
defer mon.Task()(&ctx)(&err)
if batchSize <= 0 {
@ -672,33 +672,18 @@ func (db *ProjectAccounting) ArchiveRollupsBefore(ctx context.Context, before ti
switch db.db.impl {
case dbutil.Cockroach:
for {
row := db.db.QueryRow(ctx, `
WITH rollups_to_move AS (
DELETE FROM bucket_bandwidth_rollups
WHERE interval_start <= $1
LIMIT $2 RETURNING *
), moved_rollups AS (
INSERT INTO bucket_bandwidth_rollup_archives(bucket_name, project_id, interval_start, interval_seconds, action, inline, allocated, settled)
SELECT bucket_name, project_id, interval_start, interval_seconds, action, inline, allocated, settled FROM rollups_to_move
RETURNING *
)
SELECT count(*) FROM moved_rollups
`, before, batchSize)
var rowCount int
err = row.Scan(&rowCount)
// We operate one action at a time, because we have an index on `(action, interval_start, project_id)`.
for action := range pb.PieceAction_name {
count, err := db.archiveRollupsBeforeByAction(ctx, action, before, batchSize)
archivedCount += count
if err != nil {
return bucketRollupsDeleted, err
}
bucketRollupsDeleted += rowCount
if rowCount < batchSize {
break
return archivedCount, Error.Wrap(err)
}
}
return archivedCount, nil
case dbutil.Postgres:
bwStatement := `
err := db.db.DB.QueryRow(ctx, `
WITH rollups_to_move AS (
DELETE FROM bucket_bandwidth_rollups
WHERE interval_start <= $1
@ -709,16 +694,39 @@ func (db *ProjectAccounting) ArchiveRollupsBefore(ctx context.Context, before ti
RETURNING *
)
SELECT count(*) FROM moved_rollups
`
row := db.db.DB.QueryRow(ctx, bwStatement, before)
var rowCount int
err = row.Scan(&rowCount)
if err != nil {
return bucketRollupsDeleted, err
}
bucketRollupsDeleted = rowCount
`, before).Scan(&archivedCount)
return archivedCount, Error.Wrap(err)
default:
return 0, nil
}
}
func (db *ProjectAccounting) archiveRollupsBeforeByAction(ctx context.Context, action int32, before time.Time, batchSize int) (archivedCount int, err error) {
defer mon.Task()(&ctx)(&err)
for {
var rowCount int
err := db.db.QueryRow(ctx, `
WITH rollups_to_move AS (
DELETE FROM bucket_bandwidth_rollups
WHERE action = $1 AND interval_start <= $2
LIMIT $3 RETURNING *
), moved_rollups AS (
INSERT INTO bucket_bandwidth_rollup_archives(bucket_name, project_id, interval_start, interval_seconds, action, inline, allocated, settled)
SELECT bucket_name, project_id, interval_start, interval_seconds, action, inline, allocated, settled FROM rollups_to_move
RETURNING *
)
SELECT count(*) FROM moved_rollups
`, int(action), before, batchSize).Scan(&rowCount)
if err != nil {
return archivedCount, Error.Wrap(err)
}
archivedCount += rowCount
if rowCount < batchSize {
return archivedCount, nil
}
}
return bucketRollupsDeleted, err
}
// getBucketsSinceAndBefore lists distinct bucket names for a project within a specific timeframe.