Files
zitadel/cmd/mirror/event_store.go
T
3ff4fbc7e4 fix(eventstore): add owners to unique constraints (#12740)
<!-- CURSOR_AGENT_PR_BODY_BEGIN -->
# Which Problems Are Solved

Bulk unique-constraint deletes (org, IDP with many links, user, project)
reconstruct every uniqueness key in Go and emit N `DELETE`s. That can
OOM on large tenants.

A one-shot owner backfill is also not enough: during a rolling deploy,
older binaries still insert `owners = '{}'`, so a later owner-only
delete would miss those rows.

Owner-only org/user/grant delete only matches rows whose `owners` bag
contains that tag. Several grant and member types stamped the aggregate
resource owner, which is not always the user’s home org or the granted
org. Username-scope rewrites also dropped `user:` tags, so owner-only
user delete could miss rewritten username rows.

Command must not switch to owner-only delete on the first 79 of a
version, because leftover `'{}'` rows from rolling-update overlap are
only backfilled on a later 79. A failed 79 must not turn owner-delete
on, and backfill joins must not stamp the wrong org (unverified domains,
org-scoped usernames colliding with instance-scoped ones).

# How the Problems Are Solved

Uniqueness stays the existing primary key `(instance_id, unique_type,
unique_field)`. `owners` is only a lifecycle bag of tags (`org:<id>`,
`user:<id>`, `idp:<id>`, `project:<id>`, `grant:<id>`).

- Add `owners TEXT[] NOT NULL DEFAULT '{}'` on
`eventstore.unique_constraints` (init SQL for greenfield, setup 78 for
existing clusters).
- Create `unique_constraints_owners_gin` with `CREATE INDEX
CONCURRENTLY` **outside a transaction** (same pattern as step 76).
Built-in `GIN (owners)` only; no `btree_gin`.
- Stamp `WithOwners` on **add** events. Field-level removes ignore
owners. `RemoveByOwner` stores the tag in `Owners` and both v1/v3 and v2
pushers run `DELETE … WHERE instance_id = $1 AND owners @>
ARRAY[$tag]::text[]`.
- Setup 79 backfills `owners` from projections via numbered
`cmd/setup/79/*.sql` files. Table names are interpolated from
`internal/query/projection` constants. Matches use `unique_field IN (x,
lower(x))` so the primary key can be used. Missing projection tables
(`42P01`) are skipped.
- Org domains are tagged only when `is_verified`. Instance-scoped
usernames join only users that are not org-scoped (organization settings
or domain policy `user_login_must_be_domain`). Every username row still
gets `org:` + `user:` of that user.
- 79 is a **repeatable** migration. `Check` re-runs when
`lastRun["version"]` changed or when setup-only `ForceFinalize` is set
and `lastRun["finalized"]` is still false. `Finalized` is written only
after Execute succeeds.
- Failed setup events keep the last successful `lastRun`. Command reads
`finalized` only from `system.migration.repeatable.done`.
- Org / user / project remove skip N-key event replay **when that
`finalized` bit is true**. Otherwise they keep today’s N-key lists
**and** still emit `RemoveByOwner`. The “ready” result is cached in
memory once true.

Setup writes `finalized` on the 79 repeatable.done payload:

- First 79 of a version: backfill only, `finalized=false` (unless
`ForceFinalize` is already set).
- Later 79: version change or
`BackfillUniqueConstraintOwners.ForceFinalize` /
`ZITADEL_BACKFILLUNIQUECONSTRAINTOWNERS_FORCEFINALIZE` after the rolling
update is done. That run backfills leftovers and writes
`finalized=true`.

Member and grant owner bags now include every org that should delete the
row:

- Instance, org, and project members also stamp
`org:<user_resource_owner>`.
- User grants stamp grant, user, project, and granted orgs.
- Project grant members stamp the project org, user org, and granted
org.
- Project grants stamp both orgs plus `grant:<id>`. Grant remove emits
`RemoveByOwner(grant)`.
- Username-scope rewrites keep `user:<id>` as well as `org:`.
- User/grant orgs are carried as unexported add-event fields (not
persisted JSON). `WithOwners` skips empty and duplicate tags.

# Additional Changes

- Table-driven owner-tag tests, including project grant (two `org:` tags
plus `grant:`), cross-org members, user grants, and username-scope
rewrite `user:` tags.
- Table-driven 79 `Check` + `finalizeAfterSuccess` tests for version
change, sequential version bumps, ForceFinalize restart, and
already-finalized skip.
- Command tests: a failed 79 event with `finalized=true` does not enable
owner-delete.
- Integration fixtures (verified vs unverified org domain;
instance-scoped `bob` vs org-scoped `bob`).
- Instance-only and global unique types stay untagged (`'{}'`). Instance
wipe still uses `UniqueConstraintInstanceRemove`.
- Pusher owner cases live in `eventstore_pusher_owners_test.go` so
`eventstore_pusher_test.go` stays under 1k lines.

# Additional Context

A later cleanup can delete the N-key replay helpers once a release that
ran 79 a second time (or an operator who set `ForceFinalize`) is in
production. That is not required for correctness.

<!-- CURSOR_AGENT_PR_BODY_END -->

<div><a
href="https://cursor.com/agents/bc-44f9bc9b-9a93-4cd9-8dd1-663a7b9aad2a?cursor_ref=pr_footer&cursor_cta=open_in_web"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://cursor.com/assets/images/open-in-web-dark.png"><source
media="(prefers-color-scheme: light)"
srcset="https://cursor.com/assets/images/open-in-web-light.png"><img
alt="Open in Web" width="114" height="28"
src="https://cursor.com/assets/images/open-in-web-dark.png"></picture></a>&nbsp;<a
href="https://cursor.com/background-agent?bcId=bc-44f9bc9b-9a93-4cd9-8dd1-663a7b9aad2a&cursor_ref=pr_footer&cursor_cta=open_in_cursor"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://cursor.com/assets/images/open-in-cursor-dark.png"><source
media="(prefers-color-scheme: light)"
srcset="https://cursor.com/assets/images/open-in-cursor-light.png"><img
alt="Open in Cursor" width="131" height="28"
src="https://cursor.com/assets/images/open-in-cursor-dark.png"></picture></a>&nbsp;</div>

---------

Co-authored-by: Cursor Agent <[email protected]>
Co-authored-by: Silvan <[email protected]>
2026-09-18 15:08:42 +02:00

280 lines
9.0 KiB
Go

package mirror
import (
"context"
"database/sql"
_ "embed"
"errors"
"io"
"time"
"github.com/jackc/pgx/v5/pgconn"
"github.com/jackc/pgx/v5/stdlib"
"github.com/shopspring/decimal"
"github.com/spf13/cobra"
"github.com/spf13/viper"
"github.com/zitadel/zitadel/backend/v3/instrumentation/logging"
db "github.com/zitadel/zitadel/internal/database"
"github.com/zitadel/zitadel/internal/database/dialect"
"github.com/zitadel/zitadel/internal/id"
"github.com/zitadel/zitadel/internal/v2/database"
"github.com/zitadel/zitadel/internal/v2/eventstore"
"github.com/zitadel/zitadel/internal/v2/eventstore/postgres"
"github.com/zitadel/zitadel/internal/zerrors"
)
var shouldIgnorePrevious bool
func eventstoreCmd() *cobra.Command {
cmd := &cobra.Command{
Use: "eventstore",
Short: "mirrors the eventstore of an instance from one database to another",
Long: `mirrors the eventstore of an instance from one database to another
ZITADEL needs to be initialized and set up with the --for-mirror flag
Migrate only copies events2 and unique constraints`,
RunE: func(cmd *cobra.Command, args []string) (err error) {
defer func() {
logging.OnError(cmd.Context(), err).Error("zitadel mirror eventstore command failed")
}()
config, shutdown, err := newMigrationConfig(cmd, viper.GetViper())
if err != nil {
return err
}
defer func() {
err = errors.Join(err, shutdown(cmd.Context()))
}()
copyEventstore(cmd.Context(), config)
return nil
},
}
cmd.Flags().BoolVar(&shouldReplace, "replace", false, "allow delete unique constraints of defined instances before copy")
cmd.Flags().BoolVar(&shouldIgnorePrevious, "ignore-previous", false, "ignores previous migrations of the events table")
return cmd
}
func copyEventstore(ctx context.Context, config *Migration) {
sourceClient, err := db.Connect(config.Source, false)
logging.OnError(ctx, err).Fatal("unable to connect to source database")
defer sourceClient.Close()
destClient, err := db.Connect(config.Destination, false)
logging.OnError(ctx, err).Fatal("unable to connect to destination database")
defer destClient.Close()
copyEvents(ctx, sourceClient, destClient, config.EventBulkSize)
copyUniqueConstraints(ctx, sourceClient, destClient)
}
func positionQuery(db *db.DB) (string, error) {
switch db.Type() {
case dialect.DatabaseTypePostgres:
return "SELECT EXTRACT(EPOCH FROM clock_timestamp())", nil
case dialect.DatabaseTypeCockroach:
return "SELECT cluster_logical_timestamp()", nil
default:
return "", errors.New("database type not recognized")
}
}
func copyEvents(ctx context.Context, source, dest *db.DB, bulkSize uint32) {
logging.Info(ctx, "starting to copy events")
start := time.Now()
reader, writer := io.Pipe()
migrationID, err := id.SonyFlakeGenerator().Next()
logging.OnError(ctx, err).Fatal("unable to generate migration id")
sourceConn, err := source.Conn(ctx)
logging.OnError(ctx, err).Fatal("unable to acquire source connection")
destConn, err := dest.Conn(ctx)
logging.OnError(ctx, err).Fatal("unable to acquire dest connection")
destinationES := eventstore.NewEventstoreFromOne(postgres.New(dest, &postgres.Config{
MaxRetries: 3,
}))
previousMigration, err := queryLastSuccessfulMigration(ctx, destinationES, source.DatabaseName())
logging.OnError(ctx, err).Fatal("unable to query latest successful migration")
var maxPosition decimal.Decimal
err = source.QueryRowContext(ctx,
func(row *sql.Row) error {
return row.Scan(&maxPosition)
},
"SELECT MAX(position) FROM eventstore.events2 "+instanceClause(),
)
logging.OnError(ctx, err).Fatal("unable to query max position from source")
logging.Info(ctx, "start event migration", "from", previousMigration.Position, "to", maxPosition)
nextPos := make(chan bool, 1)
pos := make(chan decimal.Decimal, 1)
errs := make(chan error, 3)
go func() {
err := sourceConn.Raw(func(driverConn interface{}) error {
conn := driverConn.(*stdlib.Conn).Conn()
nextPos <- true
var i uint32
for position := range pos {
var stmt database.Statement
stmt.WriteString("COPY (SELECT instance_id, aggregate_type, aggregate_id, event_type, sequence, revision, created_at, regexp_replace(payload::TEXT, '\\\\u0000', '', 'g')::JSON payload, creator, owner, ")
stmt.WriteArg(position)
stmt.WriteString(" position, row_number() OVER (PARTITION BY instance_id ORDER BY position, in_tx_order) AS in_tx_order FROM eventstore.events2 ")
stmt.WriteString(instanceClause())
stmt.WriteString(" AND ")
database.NewNumberAtMost(maxPosition).Write(&stmt, "position")
stmt.WriteString(" AND ")
database.NewNumberGreater(previousMigration.Position).Write(&stmt, "position")
stmt.WriteString(" ORDER BY instance_id, position, in_tx_order")
stmt.WriteString(" LIMIT ")
stmt.WriteArg(bulkSize)
stmt.WriteString(" OFFSET ")
stmt.WriteArg(bulkSize * i)
stmt.WriteString(") TO STDOUT")
// Copy does not allow args so we use we replace the args in the statement
tag, err := conn.PgConn().CopyTo(ctx, writer, stmt.Debug())
if err != nil {
return zerrors.ThrowUnknownf(err, "MIGRA-KTuSq", "unable to copy events from source during iteration %d", i)
}
logging.Info(ctx, "batch of events copied", "batch_count", i)
if tag.RowsAffected() < int64(bulkSize) {
logging.Info(ctx, "last batch of events copied", "batch_count", i)
return nil
}
nextPos <- true
i++
}
return nil
})
writer.Close()
close(nextPos)
errs <- err
}()
// generate next position for
go func() {
defer close(pos)
for range nextPos {
var position decimal.Decimal
query, err := positionQuery(dest)
if err != nil {
errs <- zerrors.ThrowUnknown(err, "MIGRA-Hy6t3", "unable to generate position query")
return
}
err = dest.QueryRowContext(
ctx,
func(row *sql.Row) error {
return row.Scan(&position)
},
query,
)
if err != nil {
errs <- zerrors.ThrowUnknown(err, "MIGRA-kMyPH", "unable to query next position")
return
}
pos <- position
}
}()
var eventCount int64
errs <- destConn.Raw(func(driverConn interface{}) error {
conn := driverConn.(*stdlib.Conn).Conn()
tag, err := conn.PgConn().CopyFrom(ctx, reader, "COPY eventstore.events2 FROM STDIN")
eventCount = tag.RowsAffected()
if err != nil {
pgErr := new(pgconn.PgError)
errors.As(err, &pgErr)
logging.WithError(ctx, err).Error("unable to copy events into destination", "pg_err_details", pgErr.Detail)
return zerrors.ThrowUnknown(err, "MIGRA-DTHi7", "unable to copy events into destination")
}
return nil
})
close(errs)
writeCopyEventsDone(ctx, destinationES, migrationID, source.DatabaseName(), maxPosition, errs)
logging.Info(ctx, "events migrated", "took", time.Since(start), "count", eventCount)
}
func writeCopyEventsDone(ctx context.Context, es *eventstore.EventStore, id, source string, position decimal.Decimal, errs <-chan error) {
joinedErrs := make([]error, 0, len(errs))
for err := range errs {
joinedErrs = append(joinedErrs, err)
}
err := errors.Join(joinedErrs...)
if err != nil {
logging.WithError(ctx, err).Error("unable to mirror events")
err := writeMigrationFailed(ctx, es, id, source, err)
logging.OnError(ctx, err).Fatal("unable to write failed event")
return
}
err = writeMigrationSucceeded(ctx, es, id, source, position)
logging.OnError(ctx, err).Fatal("unable to write succeeded event")
}
func copyUniqueConstraints(ctx context.Context, source, dest *db.DB) {
logging.Info(ctx, "starting to copy unique constraints")
start := time.Now()
reader, writer := io.Pipe()
errs := make(chan error, 1)
sourceConn, err := source.Conn(ctx)
logging.OnError(ctx, err).Fatal("unable to acquire source connection")
go func() {
err := sourceConn.Raw(func(driverConn interface{}) error {
conn := driverConn.(*stdlib.Conn).Conn()
var stmt database.Statement
stmt.WriteString("COPY (SELECT instance_id, unique_type, unique_field, owners FROM eventstore.unique_constraints ")
stmt.WriteString(instanceClause())
stmt.WriteString(") TO stdout")
_, err := conn.PgConn().CopyTo(ctx, writer, stmt.String())
writer.Close()
return err
})
errs <- err
}()
destConn, err := dest.Conn(ctx)
logging.OnError(ctx, err).Fatal("unable to acquire dest connection")
var eventCount int64
err = destConn.Raw(func(driverConn interface{}) error {
conn := driverConn.(*stdlib.Conn).Conn()
if shouldReplace {
var stmt database.Statement
stmt.WriteString("DELETE FROM eventstore.unique_constraints ")
stmt.WriteString(instanceClause())
_, err := conn.Exec(ctx, stmt.String())
if err != nil {
return err
}
}
tag, err := conn.PgConn().CopyFrom(ctx, reader, "COPY eventstore.unique_constraints (instance_id, unique_type, unique_field, owners) FROM stdin")
eventCount = tag.RowsAffected()
return err
})
logging.OnError(ctx, err).Fatal("unable to copy unique constraints to destination")
logging.OnError(ctx, <-errs).Fatal("unable to copy unique constraints from source")
logging.Info(ctx, "unique constraints migrated", "took", time.Since(start), "count", eventCount)
}