mirror of
https://github.com/mattermost/mattermost.git
synced 2025-02-25 18:55:24 -06:00
4286456077
* MM-10856: deduplicate posts with the same pending post id Leverage a fixed size cache with a window of 30 seconds to deduplicate posts received by a single app server. Clients that duplicate the same pending post id will see one request potentially delayed until the first finishes, after which the same payload should be returned by both. Duplicate posts outside the 30 second window will not be de-duplicated. Note that the cache is not synchronized between app servers. In an HA cluster consisting of more than one app server, sticky load balancing (e.g. hashing by IP or Session ID) is required to route the users to the same app instance for multiple requests. Other options considered for this feature: * adding a column to the `Posts` table: rejected as being too heavyweight * maintaining a `PendingPostIds` table: similarly rejected for the database impact * using the pending post id as the post id and relying on the unique constraints on the Post table: rejected for being difficult to show that it's safe to use a client-provided value as the row identifier * utils/lru: simplify to ttl internally and for new methods * move seenPendingPostIdsCache to App.Server * just fail concurrent post requests, vs. trying to wait * add debug log when create post is deduplicated
132 lines
3.7 KiB
Go
132 lines
3.7 KiB
Go
// Copyright (c) 2016-present Mattermost, Inc. All Rights Reserved.
|
|
// See License.txt for license information.
|
|
|
|
package store
|
|
|
|
import (
|
|
"context"
|
|
|
|
"github.com/mattermost/mattermost-server/einterfaces"
|
|
"github.com/mattermost/mattermost-server/model"
|
|
"github.com/mattermost/mattermost-server/utils"
|
|
)
|
|
|
|
const (
|
|
REACTION_CACHE_SIZE = 20000
|
|
REACTION_CACHE_SEC = 30 * 60
|
|
|
|
ROLE_CACHE_SIZE = 20000
|
|
ROLE_CACHE_SEC = 30 * 60
|
|
|
|
SCHEME_CACHE_SIZE = 20000
|
|
SCHEME_CACHE_SEC = 30 * 60
|
|
|
|
CLEAR_CACHE_MESSAGE_DATA = ""
|
|
)
|
|
|
|
type LocalCacheSupplier struct {
|
|
next LayeredStoreSupplier
|
|
reactionCache *utils.Cache
|
|
roleCache *utils.Cache
|
|
schemeCache *utils.Cache
|
|
metrics einterfaces.MetricsInterface
|
|
cluster einterfaces.ClusterInterface
|
|
}
|
|
|
|
// Caching Interface
|
|
type ObjectCache interface {
|
|
AddWithExpiresInSecs(key, value interface{}, expireAtSecs int64)
|
|
AddWithDefaultExpires(key, value interface{})
|
|
Purge()
|
|
Get(key interface{}) (value interface{}, ok bool)
|
|
Remove(key interface{})
|
|
Len() int
|
|
Name() string
|
|
GetInvalidateClusterEvent() string
|
|
}
|
|
|
|
func NewLocalCacheSupplier(metrics einterfaces.MetricsInterface, cluster einterfaces.ClusterInterface) *LocalCacheSupplier {
|
|
supplier := &LocalCacheSupplier{
|
|
reactionCache: utils.NewLruWithParams(REACTION_CACHE_SIZE, "Reaction", REACTION_CACHE_SEC, model.CLUSTER_EVENT_INVALIDATE_CACHE_FOR_REACTIONS),
|
|
roleCache: utils.NewLruWithParams(ROLE_CACHE_SIZE, "Role", ROLE_CACHE_SEC, model.CLUSTER_EVENT_INVALIDATE_CACHE_FOR_ROLES),
|
|
schemeCache: utils.NewLruWithParams(SCHEME_CACHE_SIZE, "Scheme", SCHEME_CACHE_SEC, model.CLUSTER_EVENT_INVALIDATE_CACHE_FOR_SCHEMES),
|
|
metrics: metrics,
|
|
cluster: cluster,
|
|
}
|
|
|
|
if cluster != nil {
|
|
cluster.RegisterClusterMessageHandler(model.CLUSTER_EVENT_INVALIDATE_CACHE_FOR_REACTIONS, supplier.handleClusterInvalidateReaction)
|
|
cluster.RegisterClusterMessageHandler(model.CLUSTER_EVENT_INVALIDATE_CACHE_FOR_ROLES, supplier.handleClusterInvalidateRole)
|
|
}
|
|
|
|
return supplier
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) SetChainNext(next LayeredStoreSupplier) {
|
|
s.next = next
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) Next() LayeredStoreSupplier {
|
|
return s.next
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) doStandardReadCache(ctx context.Context, cache ObjectCache, key string, hints ...LayeredStoreHint) *LayeredStoreSupplierResult {
|
|
if hintsContains(hints, LSH_NO_CACHE) {
|
|
if s.metrics != nil {
|
|
s.metrics.IncrementMemCacheMissCounter(cache.Name())
|
|
}
|
|
return nil
|
|
}
|
|
|
|
if cacheItem, ok := cache.Get(key); ok {
|
|
if s.metrics != nil {
|
|
s.metrics.IncrementMemCacheHitCounter(cache.Name())
|
|
}
|
|
result := NewSupplierResult()
|
|
result.Data = cacheItem
|
|
return result
|
|
}
|
|
|
|
if s.metrics != nil {
|
|
s.metrics.IncrementMemCacheMissCounter(cache.Name())
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) doStandardAddToCache(ctx context.Context, cache ObjectCache, key string, result *LayeredStoreSupplierResult, hints ...LayeredStoreHint) {
|
|
if result.Err == nil && result.Data != nil {
|
|
cache.AddWithDefaultExpires(key, result.Data)
|
|
}
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) doInvalidateCacheCluster(cache ObjectCache, key string) {
|
|
cache.Remove(key)
|
|
if s.cluster != nil {
|
|
msg := &model.ClusterMessage{
|
|
Event: cache.GetInvalidateClusterEvent(),
|
|
SendType: model.CLUSTER_SEND_BEST_EFFORT,
|
|
Data: key,
|
|
}
|
|
s.cluster.SendClusterMessage(msg)
|
|
}
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) doClearCacheCluster(cache ObjectCache) {
|
|
cache.Purge()
|
|
if s.cluster != nil {
|
|
msg := &model.ClusterMessage{
|
|
Event: cache.GetInvalidateClusterEvent(),
|
|
SendType: model.CLUSTER_SEND_BEST_EFFORT,
|
|
Data: CLEAR_CACHE_MESSAGE_DATA,
|
|
}
|
|
s.cluster.SendClusterMessage(msg)
|
|
}
|
|
}
|
|
|
|
func (s *LocalCacheSupplier) Invalidate() {
|
|
s.doClearCacheCluster(s.reactionCache)
|
|
s.doClearCacheCluster(s.roleCache)
|
|
s.doClearCacheCluster(s.schemeCache)
|
|
}
|