Files
zitadel/internal/eventstore/repository/search_query.go
T
4bc484dac9 perf(eventstore): read projection events per event type (#12753)
# Which Problems Are Solved

- A projection that subscribes to several event types gets slower the
further it falls behind.
- The handler reads its events with one query: `aggregate_type =
ANY(...) AND event_type = ANY(...)`, ordered by position and limited to
the bulk limit.
- The `es_projection` index returns each (aggregate type, event type)
range in position order, but Postgres cannot merge those ranges.
- So each batch reads every matching event after the projection's
position, sorts them all, and keeps only the bulk limit.
- A projection that is N events behind therefore reads about N rows per
batch. Catching up costs about N² in total, and once events arrive
faster than such batches can process them, the projection does not
recover.
- Every query that triggers the projection waits for that catch-up.
Examples are `GetSession` and the notification handlers.

# How the Problems Are Solved

- **New opt-in, `SearchQueryBuilder.ScanEventTypesSeparately()`.** It
applies to event queries for a single instance, ascending and limited,
whose sub-queries filter only by aggregate and event types.
- For such a query, the SQL layer reads each distinct (aggregate type,
event type) combination in its own `ORDER BY … LIMIT` subquery, using
`UNNEST … CROSS JOIN LATERAL`.
  - It then merges the results with the same sort key and limit.
- Each subquery is one `es_projection` range read in position order, so
it stops at the limit. A batch reads at most `limit × combinations`
rows, however far behind the projection is.
  - Every other query renders exactly as before.
- **The projection handler** now passes its real combinations, one
sub-query per aggregate type instead of the cross product of all types,
and sets the opt-in.
- **Results are identical to the single query.**
- The sort key (`position, in_tx_order, instance_id, aggregate_type,
aggregate_id, sequence`) is a total order, and the global top n is
contained in the union of each combination's top n.
- The cursor from #12703 and the open-transaction cap apply inside every
subquery.
- A test against a real Postgres compares both shapes from every cursor
and every position, for several limits. The test events include
positions shared by several transactions, whose `in_tx_order` restarts.

# Additional Changes

This is a separate commit and can be split into its own PR:
- `cmd/defaults.yaml` sets `BulkLimit: 1000` for the `sessions`
projection.
- `projection_state_latency` gets histogram buckets up to one week.
Until now, every projection more than 30 minutes behind was reported in
the same top bucket.

# Additional Context

These numbers come from a local, CPU-pinned sandbox, for comparing
builds only.
- **Database:** Postgres 17 in a VM with 8 vCPUs and 16 GiB. The
`Eventstore.Autovacuum` and `Projections` settings mirror a
production-like setup.
- **Starting state:** `sessions8` 4.8M events behind, restored
byte-identically for every run.

Batch query for `sessions8`, `EXPLAIN ANALYZE`, median of 5 runs,
parallel workers at the default:

| | before | after |
| --- | ---: | ---: |
| at the backlog, limit 1000 | 832 ms | 4.0 ms |
| at the backlog, limit 200 | 832 ms | 1.0 ms |
| at the head (idle poll) | 0.14 ms | 0.19 ms |
| generic plan (`plan_cache_mode=force_generic_plan`), at the backlog |
1,042 ms | 4.0 ms |

Draining `sessions8` from that backlog for 600 s, 3 runs per build
(median, with the range in brackets):

| | before | after |
| --- | ---: | ---: |
| events/s | 975 (932-1,032) | 7,811 (7,523-7,955) |
| mean batch time, including writes | 1.03 s | 0.13 s |
| still behind after 600 s | 4.2M | 17k-277k |

A new index `(instance_id, position, in_tx_order, aggregate_type,
event_type)` with the old query was evaluated as an alternative and
rejected. Under a generic plan, Postgres chose it for projections whose
events are rare and walked the whole event stream: 430 ms instead of
0.15 ms. This is the behaviour that led to dropping
`es_instance_position` in #10626.

- Follow-up for #12703

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5 <[email protected]>
Co-authored-by: Copilot Autofix powered by AI <[email protected]>
2026-09-22 17:09:42 +02:00

372 lines
12 KiB
Go

package repository
import (
"cmp"
"database/sql"
"slices"
"github.com/shopspring/decimal"
"github.com/zitadel/zitadel/internal/database"
"github.com/zitadel/zitadel/internal/eventstore"
"github.com/zitadel/zitadel/internal/zerrors"
)
// SearchQuery defines the which and how data are queried
type SearchQuery struct {
Columns eventstore.Columns
SubQueries [][]*Filter
Tx *sql.Tx
AwaitOpenTransactions bool
Limit uint64
Offset uint32
Desc bool
InstanceID *Filter
InstanceIDs *Filter
ExcludedInstances *Filter
Creator *Filter
Owner *Filter
Position *Filter
Sequence *Filter
CreatedAfter *Filter
CreatedBefore *Filter
ExcludeAggregateIDs []*Filter
EventSortKeyAfter *eventstore.EventSortKey
// EventTypeScans contains the combinations of aggregate type and event type to read separately,
// see [eventstore.SearchQueryBuilder.ScanEventTypesSeparately].
// It is empty if the sub queries cannot be read that way.
EventTypeScans []EventTypeScan
}
// EventTypeScan is a combination of aggregate type and event type, which is read separately.
type EventTypeScan struct {
AggregateType eventstore.AggregateType
EventType eventstore.EventType
}
// Filter represents all fields needed to compare a field of an event with a value
type Filter struct {
Field Field
Value interface{}
Operation Operation
}
// Operation defines how fields are compared
type Operation int32
const (
// OperationEquals compares two values for equality
OperationEquals Operation = iota + 1
// OperationGreater compares if the given values is greater than the stored one
OperationGreater
// OperationLess compares if the given values is less than the stored one
OperationLess
// OperationIn checks if a stored value matches one of the passed value list
OperationIn
// OperationJSONContains checks if a stored value matches the given json
OperationJSONContains
// OperationNotIn checks if a stored value does not match one of the passed value list
OperationNotIn
OperationGreaterOrEquals
operationCount
)
// Field is the representation of a field from the event
type Field int32
const (
// FieldAggregateType represents the aggregate type field
FieldAggregateType Field = iota + 1
// FieldAggregateID represents the aggregate id field
FieldAggregateID
// FieldSequence represents the sequence field
FieldSequence
// FieldResourceOwner represents the resource owner field
FieldResourceOwner
// FieldInstanceID represents the instance id field
FieldInstanceID
// FieldEditorService represents the editor service field
FieldEditorService
// FieldEditorUser represents the editor user field
FieldEditorUser
// FieldEventType represents the event type field
FieldEventType
// FieldEventData represents the event data field
FieldEventData
// FieldCreationDate represents the creation date field
FieldCreationDate
// FieldPosition represents the field of the global sequence
FieldPosition
fieldCount
)
// NewFilter is used in tests. Use searchQuery.*Filter() instead
func NewFilter(field Field, value interface{}, operation Operation) *Filter {
return &Filter{
Field: field,
Value: value,
Operation: operation,
}
}
// Validate checks if the fields of the filter have valid values
func (f *Filter) Validate() error {
if f == nil {
return zerrors.ThrowPreconditionFailed(nil, "REPO-z6KcG", "filter is nil")
}
if f.Field <= 0 || f.Field >= fieldCount {
return zerrors.ThrowPreconditionFailed(nil, "REPO-zw62U", "field not definded")
}
if f.Value == nil {
return zerrors.ThrowPreconditionFailed(nil, "REPO-GJ9ct", "no value definded")
}
if f.Operation <= 0 || f.Operation >= operationCount {
return zerrors.ThrowPreconditionFailed(nil, "REPO-RrQTy", "operation not definded")
}
return nil
}
func QueryFromBuilder(builder *eventstore.SearchQueryBuilder) (*SearchQuery, error) {
if builder == nil ||
builder.GetColumns().Validate() != nil {
return nil, zerrors.ThrowPreconditionFailed(nil, "MODEL-4m9gs", "builder invalid")
}
query := &SearchQuery{
Columns: builder.GetColumns(),
Limit: builder.GetLimit(),
Offset: builder.GetOffset(),
Desc: builder.GetDesc(),
Tx: builder.GetTx(),
AwaitOpenTransactions: builder.GetAwaitOpenTransactions(),
SubQueries: make([][]*Filter, len(builder.GetQueries())),
EventSortKeyAfter: builder.GetEventSortKeyAfter(),
}
var err error
query.EventTypeScans, err = eventTypeScans(builder)
if err != nil {
return nil, err
}
for _, f := range []func(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter{
instanceIDFilter,
instanceIDsFilter,
editorUserFilter,
resourceOwnerFilter,
positionAfterFilter,
eventSequenceGreaterFilter,
creationDateAfterFilter,
creationDateBeforeFilter,
} {
filter := f(builder, query)
if filter == nil {
continue
}
if err := filter.Validate(); err != nil {
return nil, err
}
}
for i, q := range builder.GetQueries() {
for _, f := range []func(query *eventstore.SearchQuery) *Filter{
aggregateTypeFilter,
aggregateIDFilter,
eventTypeFilter,
eventDataFilter,
eventPositionAfterFilter,
} {
filter := f(q)
if filter == nil {
continue
}
if err := filter.Validate(); err != nil {
return nil, err
}
query.SubQueries[i] = append(query.SubQueries[i], filter)
}
}
if excludeAggregateIDs := builder.GetExcludeAggregateIDs(); excludeAggregateIDs != nil {
for _, f := range []func(query *eventstore.ExclusionQuery) *Filter{
excludeAggregateTypeFilter,
excludeEventTypeFilter,
} {
filter := f(excludeAggregateIDs)
if filter == nil {
continue
}
if err := filter.Validate(); err != nil {
return nil, err
}
query.ExcludeAggregateIDs = append(query.ExcludeAggregateIDs, filter)
}
}
return query, nil
}
// eventTypeScans returns the distinct combinations of aggregate type and event type of all sub queries,
// if [eventstore.SearchQueryBuilder.ScanEventTypesSeparately] was set.
// Each sub query must filter a single aggregate type and at least one event type, and nothing else,
// otherwise an error is returned. A single aggregate type is required because event types belong to an aggregate type:
// a sub query with several aggregate types would combine event types with aggregate types they do not belong to.
func eventTypeScans(builder *eventstore.SearchQueryBuilder) ([]EventTypeScan, error) {
if !builder.GetScanEventTypesSeparately() {
return nil, nil
}
var scans []EventTypeScan
for _, query := range builder.GetQueries() {
if len(query.GetAggregateTypes()) != 1 ||
len(query.GetEventTypes()) == 0 ||
len(query.GetAggregateIDs()) > 0 ||
len(query.GetEventData()) > 0 ||
!query.GetPositionAfter().IsZero() {
return nil, zerrors.ThrowInvalidArgument(nil, "REPO-Ahx6e", "scanning event types separately requires sub queries with one aggregate type and event types only")
}
for _, eventType := range query.GetEventTypes() {
scans = append(scans, EventTypeScan{AggregateType: query.GetAggregateTypes()[0], EventType: eventType})
}
}
// sub queries are OR-connected, so a combination listed twice must still be read once
slices.SortFunc(scans, func(a, b EventTypeScan) int {
return cmp.Or(cmp.Compare(a.AggregateType, b.AggregateType), cmp.Compare(a.EventType, b.EventType))
})
return slices.Compact(scans), nil
}
func eventSequenceGreaterFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetEventSequenceGreater() == 0 {
return nil
}
sortOrder := OperationGreater
if builder.GetDesc() {
sortOrder = OperationLess
}
query.Sequence = NewFilter(FieldSequence, builder.GetEventSequenceGreater(), sortOrder)
return query.Sequence
}
func creationDateAfterFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetCreationDateAfter().IsZero() {
return nil
}
query.CreatedAfter = NewFilter(FieldCreationDate, builder.GetCreationDateAfter(), OperationGreater)
return query.CreatedAfter
}
func creationDateBeforeFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetCreationDateBefore().IsZero() {
return nil
}
query.CreatedBefore = NewFilter(FieldCreationDate, builder.GetCreationDateBefore(), OperationLess)
return query.CreatedBefore
}
func resourceOwnerFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetResourceOwner() == "" {
return nil
}
query.Owner = NewFilter(FieldResourceOwner, builder.GetResourceOwner(), OperationEquals)
return query.Owner
}
func editorUserFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetEditorUser() == "" {
return nil
}
query.Creator = NewFilter(FieldEditorUser, builder.GetEditorUser(), OperationEquals)
return query.Creator
}
func instanceIDFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetInstanceID() == nil {
return nil
}
query.InstanceID = NewFilter(FieldInstanceID, *builder.GetInstanceID(), OperationEquals)
return query.InstanceID
}
func instanceIDsFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetInstanceIDs() == nil {
return nil
}
query.InstanceIDs = NewFilter(FieldInstanceID, database.TextArray[string](builder.GetInstanceIDs()), OperationIn)
return query.InstanceIDs
}
func positionAfterFilter(builder *eventstore.SearchQueryBuilder, query *SearchQuery) *Filter {
if builder.GetPositionAtLeast().IsZero() {
return nil
}
query.Position = NewFilter(FieldPosition, builder.GetPositionAtLeast(), OperationGreaterOrEquals)
return query.Position
}
func aggregateIDFilter(query *eventstore.SearchQuery) *Filter {
if len(query.GetAggregateIDs()) < 1 {
return nil
}
if len(query.GetAggregateIDs()) == 1 {
return NewFilter(FieldAggregateID, query.GetAggregateIDs()[0], OperationEquals)
}
return NewFilter(FieldAggregateID, database.TextArray[string](query.GetAggregateIDs()), OperationIn)
}
func eventTypeFilter(query *eventstore.SearchQuery) *Filter {
if len(query.GetEventTypes()) < 1 {
return nil
}
if len(query.GetEventTypes()) == 1 {
return NewFilter(FieldEventType, query.GetEventTypes()[0], OperationEquals)
}
return NewFilter(FieldEventType, database.TextArray[eventstore.EventType](query.GetEventTypes()), OperationIn)
}
func aggregateTypeFilter(query *eventstore.SearchQuery) *Filter {
if len(query.GetAggregateTypes()) < 1 {
return nil
}
if len(query.GetAggregateTypes()) == 1 {
return NewFilter(FieldAggregateType, query.GetAggregateTypes()[0], OperationEquals)
}
return NewFilter(FieldAggregateType, database.TextArray[eventstore.AggregateType](query.GetAggregateTypes()), OperationIn)
}
func eventDataFilter(query *eventstore.SearchQuery) *Filter {
if len(query.GetEventData()) == 0 {
return nil
}
return NewFilter(FieldEventData, query.GetEventData(), OperationJSONContains)
}
func eventPositionAfterFilter(query *eventstore.SearchQuery) *Filter {
if pos := query.GetPositionAfter(); !pos.Equal(decimal.Decimal{}) {
return NewFilter(FieldPosition, pos, OperationGreater)
}
return nil
}
func excludeEventTypeFilter(query *eventstore.ExclusionQuery) *Filter {
if len(query.GetEventTypes()) < 1 {
return nil
}
if len(query.GetEventTypes()) == 1 {
return NewFilter(FieldEventType, query.GetEventTypes()[0], OperationEquals)
}
return NewFilter(FieldEventType, database.TextArray[eventstore.EventType](query.GetEventTypes()), OperationIn)
}
func excludeAggregateTypeFilter(query *eventstore.ExclusionQuery) *Filter {
if len(query.GetAggregateTypes()) < 1 {
return nil
}
if len(query.GetAggregateTypes()) == 1 {
return NewFilter(FieldAggregateType, query.GetAggregateTypes()[0], OperationEquals)
}
return NewFilter(FieldAggregateType, database.TextArray[eventstore.AggregateType](query.GetAggregateTypes()), OperationIn)
}