Files
aptly/database/goleveldb/storage.go
T
Ryan Gonzalez 19a705f80d Split reflists to share their contents across snapshots
In current aptly, each repository and snapshot has its own reflist in
the database. This brings a few problems with it:

- Given a sufficiently large repositories and snapshots, these lists can
  get enormous, reaching >1MB. This is a problem for LevelDB's overall
  performance, as it tends to prefer values around the confiruged block
  size (defaults to just 4KiB).
- When you take these large repositories and snapshot them, you have a
  full, new copy of the reflist, even if only a few packages changed.
  This means that having a lot of snapshots with a few changes causes
  the database to basically be full of largely duplicate reflists.
- All the duplication also means that many of the same refs are being
  loaded repeatedly, which can cause some slowdown but, more notably,
  eats up huge amounts of memory.
- Adding on more and more new repositories and snapshots will cause the
  time and memory spent on things like cleanup and publishing to grow
  roughly linearly.

At the core, there are two problems here:

- Reflists get very big because there are just a lot of packages.
- Different reflists can tend to duplicate much of the same contents.

*Split reflists* aim at solving this by separating reflists into 64
*buckets*. Package refs are sorted into individual buckets according to
the following system:

- Take the first 3 letters of the package name, after dropping a `lib`
  prefix. (Using only the first 3 letters will cause packages with
  similar prefixes to end up in the same bucket, under the assumption
  that packages with similar names tend to be updated together.)
- Take the 64-bit xxhash of these letters. (xxhash was chosen because it
  relatively good distribution across the individual bits, which is
  important for the next step.)
- Use the first 6 bits of the hash (range [0:63]) as an index into the
  buckets.

Once refs are placed in buckets, a sha256 digest of all the refs in the
bucket is taken. These buckets are then stored in the database, split
into roughly block-sized segments, and all the repositories and
snapshots simply store an array of bucket digests.

This approach means that *repositories and snapshots can share their
reflist buckets*. If a snapshot is taken of a repository, it will have
the same contents, so its split reflist will point to the same buckets
as the base repository, and only one copy of each bucket is stored in
the database. When some packages in the repository change, only the
buckets containing those packages will be modified; all the other
buckets will remain unchanged, and thus their contents will still be
shared. Later on, when these reflists are loaded, each bucket is only
loaded once, short-cutting loaded many megabytes of data. In effect,
split reflists are essentially copy-on-write, with only the changed
buckets stored individually.

Changing the disk format means that a migration needs to take place, so
that task is moved into the database cleanup step, which will migrate
reflists over to split reflists, as well as delete any unused reflist
buckets.

All the reflist tests are also changed to additionally test out split
reflists; although the internal logic is all shared (since buckets are,
themselves, just normal reflists), some special additions are needed to
have native versions of the various reflist helper methods.

In our tests, we've observed the following improvements:

- Memory usage during publish and database cleanup, with
  `GOMEMLIMIT=2GiB`, goes down from ~3.2GiB (larger than the memory
  limit!) to ~0.7GiB, a decrease of ~4.5x.
- Database size decreases from 1.3GB to 367MB.

*In my local tests*, publish times had also decreased down to mere
seconds but the same effect wasn't observed on the server, with the
times staying around the same. My suspicions are that this is due to I/O
performance: my local system is an M1 MBP, which almost certainly has
much faster disk speeds than our DigitalOcean block volumes. Split
reflists include a side effect of requiring more random accesses from
reading all the buckets by their keys, so if your random I/O
performance is slower, it might cancel out the benefits. That being
said, even in that case, the memory usage and database size advantages
still persist.

Signed-off-by: Ryan Gonzalez <ryan.gonzalez@collabora.com>
2025-02-15 23:49:21 +01:00

191 lines
4.5 KiB
Go

package goleveldb
import (
"bytes"
"errors"
"os"
"github.com/syndtr/goleveldb/leveldb"
"github.com/syndtr/goleveldb/leveldb/util"
"github.com/aptly-dev/aptly/database"
)
type storage struct {
path string
db *leveldb.DB
}
func (s *storage) GetRecommendedMaxKVSize() int {
// The block size configured is not actually a *set* block size, but rather a
// *minimum*. LevelDB only checks if a block is full after a new key/value pair is
// written, meaning that blocks will tend to overflow a bit.
// Therefore, using the default block size as the max value size will ensure
// that a new block will only contain a single value and that the size will
// only ever be as large as around double the block size (if the block was
// nearly full before the new items were added).
return blockSize
}
// CreateTemporary creates new DB of the same type in temp dir
func (s *storage) CreateTemporary() (database.Storage, error) {
tempdir, err := os.MkdirTemp("", "aptly")
if err != nil {
return nil, err
}
db, err := internalOpen(tempdir, true)
if err != nil {
return nil, err
}
return &storage{db: db, path: tempdir}, nil
}
// Get key value from database
func (s *storage) Get(key []byte) ([]byte, error) {
value, err := s.db.Get(key, nil)
if err != nil {
if err == leveldb.ErrNotFound {
return nil, database.ErrNotFound
}
return nil, err
}
return value, nil
}
// Put saves key to database, if key has the same value in DB already, it is not saved
func (s *storage) Put(key []byte, value []byte) error {
old, err := s.db.Get(key, nil)
if err != nil {
if err != leveldb.ErrNotFound {
return err
}
} else {
if bytes.Equal(old, value) {
return nil
}
}
return s.db.Put(key, value, nil)
}
// Delete removes key from DB
func (s *storage) Delete(key []byte) error {
return s.db.Delete(key, nil)
}
// KeysByPrefix returns all keys that start with prefix
func (s *storage) KeysByPrefix(prefix []byte) [][]byte {
result := make([][]byte, 0, 20)
iterator := s.db.NewIterator(nil, nil)
defer iterator.Release()
for ok := iterator.Seek(prefix); ok && bytes.HasPrefix(iterator.Key(), prefix); ok = iterator.Next() {
key := iterator.Key()
keyc := make([]byte, len(key))
copy(keyc, key)
result = append(result, keyc)
}
return result
}
// FetchByPrefix returns all values with keys that start with prefix
func (s *storage) FetchByPrefix(prefix []byte) [][]byte {
result := make([][]byte, 0, 20)
iterator := s.db.NewIterator(nil, nil)
defer iterator.Release()
for ok := iterator.Seek(prefix); ok && bytes.HasPrefix(iterator.Key(), prefix); ok = iterator.Next() {
val := iterator.Value()
valc := make([]byte, len(val))
copy(valc, val)
result = append(result, valc)
}
return result
}
// HasPrefix checks whether it can find any key with given prefix and returns true if one exists
func (s *storage) HasPrefix(prefix []byte) bool {
iterator := s.db.NewIterator(nil, nil)
defer iterator.Release()
return iterator.Seek(prefix) && bytes.HasPrefix(iterator.Key(), prefix)
}
// ProcessByPrefix iterates through all entries where key starts with prefix and calls
// StorageProcessor on key value pair
func (s *storage) ProcessByPrefix(prefix []byte, proc database.StorageProcessor) error {
iterator := s.db.NewIterator(nil, nil)
defer iterator.Release()
for ok := iterator.Seek(prefix); ok && bytes.HasPrefix(iterator.Key(), prefix); ok = iterator.Next() {
err := proc(iterator.Key(), iterator.Value())
if err != nil {
return err
}
}
return nil
}
// Close finishes DB work
func (s *storage) Close() error {
if s.db == nil {
return nil
}
err := s.db.Close()
s.db = nil
return err
}
// Reopen tries to open (re-open) the database
func (s *storage) Open() error {
if s.db != nil {
return nil
}
var err error
s.db, err = internalOpen(s.path, false)
return err
}
// CreateBatch creates a Batch object
func (s *storage) CreateBatch() database.Batch {
return &batch{
db: s.db,
b: &leveldb.Batch{},
}
}
// OpenTransaction creates new transaction.
func (s *storage) OpenTransaction() (database.Transaction, error) {
t, err := s.db.OpenTransaction()
if err != nil {
return nil, err
}
return &transaction{t: t}, nil
}
// CompactDB compacts database by merging layers
func (s *storage) CompactDB() error {
return s.db.CompactRange(util.Range{})
}
// Drop removes all the DB files (DANGEROUS!)
func (s *storage) Drop() error {
if s.db != nil {
return errors.New("DB is still open")
}
return os.RemoveAll(s.path)
}
// Check interface
var (
_ database.Storage = &storage{}
)