group_lifecycle.go

/*
  Copyright 2026 The ARCORIS Authors

  Licensed under the Apache License, Version 2.0 (the "License");
  you may not use this file except in compliance with the License.
  You may obtain a copy of the License at

      http://www.apache.org/licenses/LICENSE-2.0

  Unless required by applicable law or agreed to in writing, software
  distributed under the License is distributed on an "AS IS" BASIS,
  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  See the License for the specific language governing permissions and
  limitations under the License.
*/

package bufferpool

import "arcoris.dev/bufferpool/internal/multierr"

const (
	// errGroupClosed reports operations rejected after group shutdown starts.
	errGroupClosed = "bufferpool.PoolGroup: group is closed"
)

// Lifecycle returns the current group lifecycle state.
func (g *PoolGroup) Lifecycle() LifecycleState { g.mustBeInitialized(); return g.lifecycle.Load() }

// IsClosed reports whether the group is closed.
func (g *PoolGroup) IsClosed() bool { g.mustBeInitialized(); return g.lifecycle.IsClosed() }

// Close performs the current hard group shutdown.
//
// Close closes every group-owned PoolPartition and aggregates close errors.
//
// Close first serializes with live PublishPolicy scheduler reconfiguration, then
// stops any opt-in coordinator scheduler outside runtimeMu so a scheduled
// TickInto cannot be blocked on runtimeMu.RLock while Close waits for scheduler
// shutdown. It then owns runtimeMu so in-flight group-routed foreground
// operations finish before Closing is published. BeginClose gates cleanup: only
// the caller that moves the group into Closing closes child partitions and
// publishes Closed. If Closing is already visible after runtimeMu is owned, no
// other group Close cleanup can still be running, so this caller may finish the
// already-started shutdown. Concurrent Close callers wait on runtimeMu, then
// return nil without duplicating child cleanup. Close is not a graceful drain
// controller, does not wait beyond partition Close behavior, and does not
// coordinate physical trim. Close updates lifecycle state; it does not rewrite
// ControllerStatus unless a later foreground TickInto observes the closed
// lifecycle and publishes Closed.
func (g *PoolGroup) Close() error {
	g.mustBeInitialized()

	g.schedulerReconfigMu.Lock()
	defer g.schedulerReconfigMu.Unlock()

	g.stopCoordinatorScheduler()

	g.runtimeMu.Lock()
	defer g.runtimeMu.Unlock()

	if !beginOrContinueSerializedCloseCleanupLocked(&g.lifecycle) {
		return nil
	}

	var err error
	if closeErr := g.registry.closeAll(); closeErr != nil {
		multierr.AppendInto(&err, closeErr)
	}

	g.lifecycle.MarkClosed()
	g.generation.Advance()

	return err
}