github.com/dgraph-io/dgraph

• Added

• Add --cache_mb and --cache_percentage flags. (#6288)

• Add flag to set up compression in zero. (#6357)

• Add flags to set table and vlog loading mode for zero. (#6344)

• Fixed

• Don't allow idx flag to be set to 0 on dgraph zero. (#6193)

• Stop forcing RAM mode for the write-ahead log. (#6261)

• Return error on closed DB. (#6319)

• Don't store start_ts in postings. (#6212)

• Optimize splits by doing binary search. Clear the pack from the main list. (#6334)

• Add a lock to backups to process one request at a time. (#6341)

• Use z.Closer instead of y.Closer' (#6396)

• Force exit if CTRL-C is caught before initialization. (#6408)

• Fix(Alpha): MASA: Make Alpha Shutdown Again. (#6406)

• Enterprise features

  • Sort manifests by backup number. (#6281)
  • Skip backing up nil lists. (#6314)

Changed

• Update Badger. (#5940, #5990)
  • Fix assert in background compression and encryption. (dgraph-io/badger#1366)
  • Avoid panic in filltables() (dgraph-io/badger#1365)
  • Force KeepL0InMemory to be true when InMemory is true (dgraph-io/badger#1375)
  • Tests: Use t.Parallel in TestIteratePrefix tests (dgraph-io/badger#1377)
  • Remove second initialization of writech in Open (dgraph-io/badger#1382)
  • Increase default valueThreshold from 32B to 1KB (dgraph-io/badger#1346)
  • Pre allocate cache key for the block cache and the bloom filter cache (dgraph-io/badger#1371)
  • Rework DB.DropPrefix (dgraph-io/badger#1381)
  • Update head while replaying value log (dgraph-io/badger#1372)
  • Update ristretto to commit f66de99 (dgraph-io/badger#1391)
  • Enable cross-compiled 32bit tests on TravisCI (dgraph-io/badger#1392)
  • Avoid panic on multiple closer.Signal calls (dgraph-io/badger#1401)
  • Add a contribution guide (dgraph-io/badger#1379)
  • Add assert to check integer overflow for table size (dgraph-io/badger#1402)
  • Return error if the vlog writes exceeds more that 4GB. (dgraph-io/badger#1400)
  • Revert "add assert to check integer overflow for table size (dgraph-io/badger#1402)" (dgraph-io/badger#1406)
  • Revert "fix: Fix race condition in block.incRef (dgraph-io/badger#1337)" (dgraph-io/badger#1407)
  • Revert "Buffer pool for decompression (dgraph-io/badger#1308)" (dgraph-io/badger#1408)
  • Revert "Compress/Encrypt Blocks in the background (dgraph-io/badger#1227)" (dgraph-io/badger#1409)
  • Add missing changelog for v2.0.3 (dgraph-io/badger#1410)
  • Changelog for v20.07.0 (dgraph-io/badger#1411)

Fixed

• Alpha: Enable bloom filter caching. (#5554)
• K shortest paths queries fix. (#5596)
• Add check on rebalance interval. (#5595)
• Change error message in case of successful license application. (#5593)
• Remove noisy logs in tablet move. (#5592)
• Avoid assigning duplicate RAFT IDs to new nodes. Fixes [#5436][]. (#5603)
• Update badger: Set KeepL0InMemory to false (badger default), and Set DetectConflicts to false. (#5615)
• Use /tmp dir to store temporary index. Fixes #4600. (#5730)
• Split posting lists recursively. (#4867)
• Set version when rollup is called with no splits. (#4945)
• Return error instead of panic (readPostingList). Fixes #5749. (#5908)
• ServeTask: Return error if server is not ready. (#6022)
• Enterprise features
  • Backup: Change groupId from int to uint32. (#5613)
  • Backup: During backup, collapse split posting lists into a single list. (#4682)
  • Backup: Use a sync.Pool to allocate KVs during backup. ([#5579][])

Changed

• Return error response if encoded response is > 4GB in size. Replace idMap with idSlice in encoder. ([#5359][])
• Change the default ratio of traces from 1 to 0.01. ([#5405][])

Fixed

• Export: Ignore deleted predicates from schema. Fixes [#5053][]. (#5327)
• Fix segmentation fault in query.go. ([#5377][])
• Update group checksums when combining multiple deltas. Fixes [#5368][]. ([#5394][])
• Fix empty string checks. (#5396)
• Fix protobuf headers check. ([#5381][])
• Stream the full set of predicates and types during a snapshot. ([#5444][])
• Use pre-allocated protobufs during backups. (#5508)
• Replace strings.Trim with strings.TrimFunc in ParseRDF. ([#5494][])
• Return nil instead of emptyTablet in groupi.Tablet(). ([#5469][])
• During shutdown, generate snapshot before closing raft node. ([#5476][])
• Get lists of predicates and types before sending the snapshot. ([#5488][])
• Move runVlogGC to x and use it in zero as well. (#5468)
• Fix inconsistent bulk loader failures. Fixes [#5361][]. ([#5537][])

Security

• Use SensitiveByteSlice type for hmac secret. (#5451)

This release was removed

Changed

• Support comma separated list of zero addresses in alpha. (#5258)
• Optimization: Optimize snapshot creation. (#4901)
• Optimization: Remove isChild from fastJsonNode. (#5184)
• Optimization: Memory improvements in fastJsonNode. (#5088)
• Update Badger to commit cddf7c03451c33. (#5273)
  • Compression/encryption runs in the background (which means faster writes)
  • Separate cache for bloom filters which limits the amount of memory used by bloom filters
• Avoid crashing live loader in case the network is interrupted. (#5268)
• Enterprise features
  • Backup/restore: Force users to explicitly tell restore command to run without zero. ([#5206][])

Fixed

• Check uid list is empty when filling shortest path vars. ([#5152][])
• Return error for invalid UID 0x0. Fixes [#5238][]. ([#5252][])
• Skipping floats that cannot be marshalled (+Inf, -Inf, NaN). ([#5199][], #5163)
• Set correct posting list type while creating it in live loader. (#5012)
• Add support for tinyint in migrate tool. Fixes #4674. (#4842)
• Fix bug, aggregate value var works with blank node in upsert. Fixes [#4712][]. ([#4767][])
• Always set BlockSize in encoder. Fixes #5102. (#5255)
• Enterprise features
  • Backup schema keys in incremental backups. Before, the schema was only stored in the full backup. ([#5158][])

Added

• Add Badger expvar metrics to Prometheus metrics. Fixes #4772. (#5094)
• Enterprise features
  • Support bulk loader use-case to import unencrypted export and encrypt. ([#5213][])
  • Create encrypted restore directory from encrypted backups. ([#5144][])
  • Add option "--encryption_key_file"/"-k" to debug tool for encryption support. ([#5146][])
  • Support for encrypted backups/restore. Note: Older backups without encryption will be incompatible with this Dgraph version. Solution is to force a full backup before creating further incremental backups. ([#5103][])
  • Add encryption support for export and import (via bulk, live loaders). ([#5155][])

Changed

• Wrap errors thrown in posting/list.go for easier debugging. (#4880)
• Print keys using hex encoding in error messages in list.go. (#4891)

Fixed

• Do not include empty nodes in the export output. (#4896)
• Fix error when lexing language list. (#4784)
• Properly initialize posting package in debug tool. ([#4893][])
• Handle special characters in schema and type queries. Fixes #4933. (#4937)
• Overwrite values for uid predicates. Fixes #4879. (#4883)
• Disable @* language queries when the predicate does not support langs. (#4881)
• Fix bug in exporting types with reverse predicates. Fixes #4856. (#4857)
• Do not skip over split keys. (Trying to skip over the split keys sometimes skips over keys belonging to a different split key. This is a fix just for this release as the actual fix requires changes to the data format.) (#4951)
• Fix point-in-time Prometheus metrics. Fixes #4532. (#4948)
• Split lists in the bulk loader. (#4967)
• Allow remote MySQL server with dgraph migrate tool. Fixes #4707. (#4860)
• Enterprise features
  • ACL: Allow uid access. (#4922)
  • Backups: Assign maxLeaseId during restore. Fixes #4816. (#4877)
  • Backups: Verify host when default and custom credentials are used. Fixes #4855. (#4858)
  • Backups: Split lists when restoring from backup. (#4912)

Fixed

• Fix bug related to posting list split, and re-enable posting list splits. Fixes #4733. (#4742)

Changed

• Allow overwriting values of predicates of type uid. Fixes #4136. (#4411)
• Algorithms to handle UidPack. (#4321)
• Improved latency in live loader using conflict resolution at client level. (#4362)
• Set ZSTD CompressionLevel to 1. (#4572)
• Splits are now disabled. (#4672)
• Disk based re-indexing: while re-indexing a predicate, the temp data is now written on disk instead of keeping it in memory. This improves index rebuild for large datasets. (#4440)
• Enterprise features
  • Breaking changes
    • Change default behavior to block operations with ACLs enabled. (#4390)
  • Remove unauthorized predicates from query instead of rejecting the query entirely. (#4479)

Added

• Add debuginfo subcommand to dgraph. (#4464)
• Support filtering on non-indexed predicate. Fixes #4305. (#4531)
• Add support for variables in recurse. Fixes #3301. (#4385).
• Adds @noconflict schema directive to prevent conflict detection. This is an experimental feature. This is not a recommended directive, but exists to help avoid conflicts for predicates which don't have high correctness requirements. Fixes #4079. (#4454)
• Implement the state HTTP endpoint on Alpha. Login is required if ACL is enabled. (#4435).
• Implement /health?all endpoint on Alpha nodes. (#4535)
• Add /health endpoint to Zero. (#4405)
• Breaking changes
  • Support for fetching facets from value edge list. The query response format is backwards-incompatible. Fixes #4081. (#4267)
• Enterprise features
  • Add guardians group with full authorization. (#4447)

Fixed

• Infer type of schema from JSON and RDF mutations. Fixes #3788. (#4328)
• Fix retrieval of facets with cascade. Fixes #4310. (#4530)
• Do not use type keys during tablet size calculation. Fixes #4473. (#4517)
• Fix Levenshtein distance calculation with match function. Fixes #4494. (#4545)
• Add <xs:integer> RDF type for int schema type. Fixes #4460. (#4465)
• Allow @filter directive with expand queries. Fixes #3904. (#4404).
• A multi-part posting list should only be accessed via the main key. Accessing the posting list via one of the other keys was causing issues during rollup and adding spurious keys to the database. Now fixed. (#4574)
• Enterprise features
  • Backup types. Fixes #4507. (#4514)