fix: limit the scope of the template average build time query to the last 100 (#19648)

This PR should resolve https://github.com/coder/internal/issues/719 by
limiting the `workspace_builds` rows selected by the query to the most
recent 100 builds of a template, as opposed to all builds in the last
30d. For our own internal templates with the most builds (1700-2000 in a
30d period) this should cut the query execution time by about 80%.

Unless we have some restriction on keeping the 30d period, contract
related or otherwise, this seems like a safe change to make. In addition
to the execution speed improvements it also means the memory for the
query is bounded as well.

If we want to keep a 30d time period for the avg build time value I
think it's worth exploring a purpose built solution such as histogram
structures where the build times could be bucketized by template ID as
they're observed.

---------

Signed-off-by: Callum Styan <callumstyan@gmail.com>
This commit is contained in:
Callum Styan
2025-09-01 09:31:21 -07:00
committed by GitHub
parent 6574299fcc
commit 4fab14b40b
8 changed files with 15 additions and 24 deletions
+1 -1
View File
@@ -2792,7 +2792,7 @@ func (q *querier) GetTemplateAppInsightsByTemplate(ctx context.Context, arg data
}
// Only used by metrics cache.
func (q *querier) GetTemplateAverageBuildTime(ctx context.Context, arg database.GetTemplateAverageBuildTimeParams) (database.GetTemplateAverageBuildTimeRow, error) {
func (q *querier) GetTemplateAverageBuildTime(ctx context.Context, arg uuid.NullUUID) (database.GetTemplateAverageBuildTimeRow, error) {
if err := q.authorizeContext(ctx, policy.ActionRead, rbac.ResourceSystem); err != nil {
return database.GetTemplateAverageBuildTimeRow{}, err
}
+1 -1
View File
@@ -3160,7 +3160,7 @@ func (s *MethodTestSuite) TestSystemFunctions() {
check.Args(arg).Asserts(rbac.ResourceSystem, policy.ActionRead).Returns(p)
}))
s.Run("GetTemplateAverageBuildTime", s.Mocked(func(dbm *dbmock.MockStore, _ *gofakeit.Faker, check *expects) {
arg := database.GetTemplateAverageBuildTimeParams{}
arg := uuid.NullUUID{}
dbm.EXPECT().GetTemplateAverageBuildTime(gomock.Any(), arg).Return(database.GetTemplateAverageBuildTimeRow{}, nil).AnyTimes()
check.Args(arg).Asserts(rbac.ResourceSystem, policy.ActionRead)
}))
+1 -1
View File
@@ -1447,7 +1447,7 @@ func (m queryMetricsStore) GetTemplateAppInsightsByTemplate(ctx context.Context,
return r0, r1
}
func (m queryMetricsStore) GetTemplateAverageBuildTime(ctx context.Context, arg database.GetTemplateAverageBuildTimeParams) (database.GetTemplateAverageBuildTimeRow, error) {
func (m queryMetricsStore) GetTemplateAverageBuildTime(ctx context.Context, arg uuid.NullUUID) (database.GetTemplateAverageBuildTimeRow, error) {
start := time.Now()
buildTime, err := m.s.GetTemplateAverageBuildTime(ctx, arg)
m.queryLatencies.WithLabelValues("GetTemplateAverageBuildTime").Observe(time.Since(start).Seconds())
+4 -4
View File
@@ -3047,18 +3047,18 @@ func (mr *MockStoreMockRecorder) GetTemplateAppInsightsByTemplate(ctx, arg any)
}
// GetTemplateAverageBuildTime mocks base method.
func (m *MockStore) GetTemplateAverageBuildTime(ctx context.Context, arg database.GetTemplateAverageBuildTimeParams) (database.GetTemplateAverageBuildTimeRow, error) {
func (m *MockStore) GetTemplateAverageBuildTime(ctx context.Context, templateID uuid.NullUUID) (database.GetTemplateAverageBuildTimeRow, error) {
m.ctrl.T.Helper()
ret := m.ctrl.Call(m, "GetTemplateAverageBuildTime", ctx, arg)
ret := m.ctrl.Call(m, "GetTemplateAverageBuildTime", ctx, templateID)
ret0, _ := ret[0].(database.GetTemplateAverageBuildTimeRow)
ret1, _ := ret[1].(error)
return ret0, ret1
}
// GetTemplateAverageBuildTime indicates an expected call of GetTemplateAverageBuildTime.
func (mr *MockStoreMockRecorder) GetTemplateAverageBuildTime(ctx, arg any) *gomock.Call {
func (mr *MockStoreMockRecorder) GetTemplateAverageBuildTime(ctx, templateID any) *gomock.Call {
mr.mock.ctrl.T.Helper()
return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetTemplateAverageBuildTime", reflect.TypeOf((*MockStore)(nil).GetTemplateAverageBuildTime), ctx, arg)
return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetTemplateAverageBuildTime", reflect.TypeOf((*MockStore)(nil).GetTemplateAverageBuildTime), ctx, templateID)
}
// GetTemplateByID mocks base method.
+1 -1
View File
@@ -325,7 +325,7 @@ type sqlcQuerier interface {
// GetTemplateAppInsightsByTemplate is used for Prometheus metrics. Keep
// in sync with GetTemplateAppInsights and UpsertTemplateUsageStats.
GetTemplateAppInsightsByTemplate(ctx context.Context, arg GetTemplateAppInsightsByTemplateParams) ([]GetTemplateAppInsightsByTemplateRow, error)
GetTemplateAverageBuildTime(ctx context.Context, arg GetTemplateAverageBuildTimeParams) (GetTemplateAverageBuildTimeRow, error)
GetTemplateAverageBuildTime(ctx context.Context, templateID uuid.NullUUID) (GetTemplateAverageBuildTimeRow, error)
GetTemplateByID(ctx context.Context, id uuid.UUID) (Template, error)
GetTemplateByOrganizationAndName(ctx context.Context, arg GetTemplateByOrganizationAndNameParams) (Template, error)
GetTemplateDAUs(ctx context.Context, arg GetTemplateDAUsParams) ([]GetTemplateDAUsRow, error)
+3 -8
View File
@@ -11903,11 +11903,11 @@ JOIN provisioner_jobs pj ON
WHERE
template_versions.template_id = $1 AND
(pj.completed_at IS NOT NULL) AND (pj.started_at IS NOT NULL) AND
(pj.started_at > $2) AND
(pj.canceled_at IS NULL) AND
((pj.error IS NULL) OR (pj.error = ''))
ORDER BY
workspace_builds.created_at DESC
LIMIT 100
)
SELECT
-- Postgres offers no clear way to DRY this short of a function or other
@@ -11921,11 +11921,6 @@ SELECT
FROM build_times
`
type GetTemplateAverageBuildTimeParams struct {
TemplateID uuid.NullUUID `db:"template_id" json:"template_id"`
StartTime sql.NullTime `db:"start_time" json:"start_time"`
}
type GetTemplateAverageBuildTimeRow struct {
Start50 float64 `db:"start_50" json:"start_50"`
Stop50 float64 `db:"stop_50" json:"stop_50"`
@@ -11935,8 +11930,8 @@ type GetTemplateAverageBuildTimeRow struct {
Delete95 float64 `db:"delete_95" json:"delete_95"`
}
func (q *sqlQuerier) GetTemplateAverageBuildTime(ctx context.Context, arg GetTemplateAverageBuildTimeParams) (GetTemplateAverageBuildTimeRow, error) {
row := q.db.QueryRowContext(ctx, getTemplateAverageBuildTime, arg.TemplateID, arg.StartTime)
func (q *sqlQuerier) GetTemplateAverageBuildTime(ctx context.Context, templateID uuid.NullUUID) (GetTemplateAverageBuildTimeRow, error) {
row := q.db.QueryRowContext(ctx, getTemplateAverageBuildTime, templateID)
var i GetTemplateAverageBuildTimeRow
err := row.Scan(
&i.Start50,
+1 -1
View File
@@ -203,11 +203,11 @@ JOIN provisioner_jobs pj ON
WHERE
template_versions.template_id = @template_id AND
(pj.completed_at IS NOT NULL) AND (pj.started_at IS NOT NULL) AND
(pj.started_at > @start_time) AND
(pj.canceled_at IS NULL) AND
((pj.error IS NULL) OR (pj.error = ''))
ORDER BY
workspace_builds.created_at DESC
LIMIT 100
)
SELECT
-- Postgres offers no clear way to DRY this short of a function or other
+3 -7
View File
@@ -101,16 +101,12 @@ func (c *Cache) refreshTemplateBuildTimes(ctx context.Context) error {
for _, template := range templates {
ids = append(ids, template.ID)
templateAvgBuildTime, err := c.database.GetTemplateAverageBuildTime(ctx, database.GetTemplateAverageBuildTimeParams{
TemplateID: uuid.NullUUID{
templateAvgBuildTime, err := c.database.GetTemplateAverageBuildTime(ctx,
uuid.NullUUID{
UUID: template.ID,
Valid: true,
},
StartTime: sql.NullTime{
Time: dbtime.Time(c.clock.Now().AddDate(0, 0, -30)),
Valid: true,
},
})
)
if err != nil {
return err
}