tile38/internal/server/expire.go

88 lines
1.7 KiB
Go
Raw Normal View History

package server
2016-07-15 22:22:48 +03:00
import (
"time"
Fix excessive memory usage for objects with TTLs This commit fixes an issue where Tile38 was using lots of extra memory to track objects that are marked to expire. This was creating problems with applications that set big TTLs. How it worked before: Every collection had a unique hashmap that stores expiration timestamps for every object in that collection. Along with the hashmaps, there's also one big server-wide list that gets appended every time a new SET+EX is performed. From a background routine, this list is looped over at least 10 times per second and is randomly searched for potential candidates that might need expiring. The routine then removes those entries from the list and tests if the objects matching the entries have actually expired. If so, these objects are deleted them from the database. When at least 25% of the 20 candidates are deleted the loop is immediately continued, otherwise the loop backs off with a 100ms pause. Why this was a problem. The list grows one entry for every SET+EX. When TTLs are long, like 24-hours or more, it would take at least that much time before the entry is removed. So for databased that have objects that use TTLs and are updated often this could lead to a very large list. How it was fixed. The list was removed and the hashmap is now search randomly. This required a new hashmap implementation, as the built-in Go map does not provide an operation for randomly geting entries. The chosen implementation is a robinhood-hash because it provides open-addressing, which makes for simple random bucket selections. Issue #502
2019-10-29 21:04:07 +03:00
"github.com/tidwall/tile38/internal/log"
2016-07-15 22:22:48 +03:00
)
Fix excessive memory usage for objects with TTLs This commit fixes an issue where Tile38 was using lots of extra memory to track objects that are marked to expire. This was creating problems with applications that set big TTLs. How it worked before: Every collection had a unique hashmap that stores expiration timestamps for every object in that collection. Along with the hashmaps, there's also one big server-wide list that gets appended every time a new SET+EX is performed. From a background routine, this list is looped over at least 10 times per second and is randomly searched for potential candidates that might need expiring. The routine then removes those entries from the list and tests if the objects matching the entries have actually expired. If so, these objects are deleted them from the database. When at least 25% of the 20 candidates are deleted the loop is immediately continued, otherwise the loop backs off with a 100ms pause. Why this was a problem. The list grows one entry for every SET+EX. When TTLs are long, like 24-hours or more, it would take at least that much time before the entry is removed. So for databased that have objects that use TTLs and are updated often this could lead to a very large list. How it was fixed. The list was removed and the hashmap is now search randomly. This required a new hashmap implementation, as the built-in Go map does not provide an operation for randomly geting entries. The chosen implementation is a robinhood-hash because it provides open-addressing, which makes for simple random bucket selections. Issue #502
2019-10-29 21:04:07 +03:00
const bgExpireDelay = time.Second / 10
// backgroundExpiring deletes expired items from the database.
// It's executes every 1/10 of a second.
func (s *Server) backgroundExpiring() {
Fix excessive memory usage for objects with TTLs This commit fixes an issue where Tile38 was using lots of extra memory to track objects that are marked to expire. This was creating problems with applications that set big TTLs. How it worked before: Every collection had a unique hashmap that stores expiration timestamps for every object in that collection. Along with the hashmaps, there's also one big server-wide list that gets appended every time a new SET+EX is performed. From a background routine, this list is looped over at least 10 times per second and is randomly searched for potential candidates that might need expiring. The routine then removes those entries from the list and tests if the objects matching the entries have actually expired. If so, these objects are deleted them from the database. When at least 25% of the 20 candidates are deleted the loop is immediately continued, otherwise the loop backs off with a 100ms pause. Why this was a problem. The list grows one entry for every SET+EX. When TTLs are long, like 24-hours or more, it would take at least that much time before the entry is removed. So for databased that have objects that use TTLs and are updated often this could lead to a very large list. How it was fixed. The list was removed and the hashmap is now search randomly. This required a new hashmap implementation, as the built-in Go map does not provide an operation for randomly geting entries. The chosen implementation is a robinhood-hash because it provides open-addressing, which makes for simple random bucket selections. Issue #502
2019-10-29 21:04:07 +03:00
for {
if s.stopServer.on() {
Fix excessive memory usage for objects with TTLs This commit fixes an issue where Tile38 was using lots of extra memory to track objects that are marked to expire. This was creating problems with applications that set big TTLs. How it worked before: Every collection had a unique hashmap that stores expiration timestamps for every object in that collection. Along with the hashmaps, there's also one big server-wide list that gets appended every time a new SET+EX is performed. From a background routine, this list is looped over at least 10 times per second and is randomly searched for potential candidates that might need expiring. The routine then removes those entries from the list and tests if the objects matching the entries have actually expired. If so, these objects are deleted them from the database. When at least 25% of the 20 candidates are deleted the loop is immediately continued, otherwise the loop backs off with a 100ms pause. Why this was a problem. The list grows one entry for every SET+EX. When TTLs are long, like 24-hours or more, it would take at least that much time before the entry is removed. So for databased that have objects that use TTLs and are updated often this could lead to a very large list. How it was fixed. The list was removed and the hashmap is now search randomly. This required a new hashmap implementation, as the built-in Go map does not provide an operation for randomly geting entries. The chosen implementation is a robinhood-hash because it provides open-addressing, which makes for simple random bucket selections. Issue #502
2019-10-29 21:04:07 +03:00
return
}
func() {
s.mu.Lock()
defer s.mu.Unlock()
2021-09-12 19:55:58 +03:00
now := time.Now()
s.backgroundExpireObjects(now)
s.backgroundExpireHooks(now)
}()
time.Sleep(bgExpireDelay)
2016-07-15 22:22:48 +03:00
}
}
2021-09-12 19:55:58 +03:00
func (s *Server) backgroundExpireObjects(now time.Time) {
nano := now.UnixNano()
var ids []string
var msgs []*Message
s.cols.Ascend(nil, func(v interface{}) bool {
col := v.(*collectionKeyContainer)
ids = col.col.Expired(nano, ids[:0])
for _, id := range ids {
msgs = append(msgs, &Message{
Args: []string{"del", col.key, id},
})
}
return true
})
for _, msg := range msgs {
_, d, err := s.cmdDel(msg)
if err != nil {
log.Fatal(err)
}
if err := s.writeAOF(msg.Args, &d); err != nil {
log.Fatal(err)
}
}
if len(msgs) > 0 {
log.Debugf("Expired %d objects\n", len(msgs))
}
}
func (s *Server) backgroundExpireHooks(now time.Time) {
var msgs []*Message
s.hookExpires.Ascend(nil, func(v interface{}) bool {
h := v.(*Hook)
if h.expires.After(now) {
return false
}
msg := &Message{}
if h.channel {
msg.Args = []string{"delchan", h.Name}
} else {
msg.Args = []string{"delhook", h.Name}
}
msgs = append(msgs, msg)
return true
})
for _, msg := range msgs {
_, d, err := s.cmdDelHook(msg)
2021-09-12 19:55:58 +03:00
if err != nil {
log.Fatal(err)
}
if err := s.writeAOF(msg.Args, &d); err != nil {
log.Fatal(err)
}
}
if len(msgs) > 0 {
log.Debugf("Expired %d hooks\n", len(msgs))
}
}