- [Samovar] Improve
wait_before_processingparameter
- [Samovar] Fix samovar without sync phase
- Fix segfault when rescan happens
- [Teapotless][Samovar] Add file limit for distributed metadata parsing
- [Teapotless][Samovar] Optimize position delete pruning
- [Teapotless][Samovar] Fix reading empty Iceberg tables when using External table
- [Samovar] Fix samovar queue name handling
- [Samovar] Do not update TTL too often
- [Samovar] Clear metadata after query completes
- [Samovar] Return metadata filtering
- [Samovar] Batch data files before pushing to Redis
- [Samovar] Ignore deleted files when using distributed metadata processing
- [Samovar] Implement distributed metadata processing for large tables
- [Samovar] Do not block the coordinator (as a writer) during the sync phase
- [Teapotless][Samovar] Improve dangling positional delete pruning
- [Samovar] Sleep before processing for large slice numbers
- [Samovar] Make initial synchronization timeouts configurable
- Fix reading float4 values in tables with float8 type
- [Samovar] Add jitter
- [Samovar] Static balancing for small tables
- [Samovar] Make initial synchronization configurable
- [Samovar] Handle cancel command when reading metadata
- Make arrow batch size adaptive
- Make arrow batch size configurable
- Set the content type to 'application/gzip' when uploading the artifact
- [Teapotless][Samovar] Skip dangling positional delete files
- Fix LIKE function for strings with
\ncharacter
- Recreate UDF in tea schema
- Fix downgrade/upgrade on gp5
- Implement UDF to get primitive metrics from Iceberg tables
- Improve reading in case filter skips all rows in row group
- [Teapotless][Samovar] Improve metadata reading for queries without filter
- [Teapotless][Samovar] Improve metadata reading for queries with partition filter
- Fix 'prefetch' mode
- Validate profile to tables mapping
- [Samovar] Remove fallback logic when it is redundant
- [Samovar] Improve reading partitioned table with deletes
- [Samovar] Log samovar metrics
- [Samovar] Improve TTL setting for keys
- Log metadata access type
- Return error in case some segments use teapot and some segments use samovar
- Support overriding profile in tea-config.json
- Skip row groups in positional deletes when possible
- Speed up metadata reading for queries without filter
- Support
coalesceexpression in filters
- [Teapot] Improve delete files reading (do not reopen them if possible)
- [Teapotless][Samovar] Support comparison of different temporal types in min/max filter
- Validate schema of config in JSON format
- Fix potential bug if 'prefetch' is enabled
- Do not support config in ini format anymore
- [Teapotless][Samovar] Improve Iceberg metadata handling (make less copies)
- [Teapotless][Samovar] Apply positional delete files according to referenced_data_file if set
- [Teapotless][Samovar] Apply metadata filter as early as possible
- [Teapot] Improve teapot error messages
- Improve error messages about types incompatibility between greenplum and iceberg
- [Teapotless][Samovar] Fix incorrect metadata reading when iceberg manifest file contains at least two entries
- [Samovar] Possibility to choose coordinator from all segments instead of segment 0
- Annotated logs with actual timestamps
- [Samovar] Fix
Int32 overflowwhen parquet files are larger than 2GB - [Samovar] Fix
Positional delete stream error - [Samovar] Fix
Extra reading from hms and s3
- [Samovar] Change default policy in splitting tasks
- [Samovar] Fix
Backoff limit exceeded
- [Teapotless] Fix logical type validation for 'day' transform
- [Teapot] Fix 'connection attempt timed out before receiving SETTINGS frame'
- [Samovar] Fix exponential backoff
- [Samovar] Abort request if connection was broken while loading task from queue
- [Teapotless][Samovar] Use partitioning info from iceberg for applying delete files
- Unify metadata access methods
- [Teapotless][Samovar] Decrease metadata size via encoding filenames
- [Teapotless][Samovar] Decrease metadata memory consumption via discarding unnecessary meta
- [Teapotless][Samovar] Measure iceberg metrics
- [Samovar] Measure samovar metrics
- [Samovar] Make samovar logs more verbose
- Fix incorrect missing RowGroup::file_offset handling
- Implement filters on iceberg metadata level
- Support config in JSON format
- Add config validator
- Make uuid work without mac address
- Improve endpoints handling in config
- Fix missing positional delete file handling
- Fix processing order for PlannedMetadata
- Do not depend on order of AnnotatedDataEntry in Reader::Advance
- Add flag for enabling postfilter on gp
- Configure access to meta via .cfg file
- Add parameter 'json_max_message_size_on_master'
- Add fallback for HMS
- Make filters more usable
- Balance reading iceberg data entries by split offsets
- Handle missing positional delete file
- Enable
use_row_filteroption for FDW by default
- Handle missing column in equality delete file
- CP1251 substitutor for Json columns
- Support row-group filter for
CASE
- Add
filter_ignored_op_exprsandfilter_ignored_func_exprsparameteres - Add
prefetchparameter
- Drop support for upgrades from Tea before 1.56.2
- Add
CREATE SERVER tea_server ...to extension SQL scripts
- Fix 'Corrupt snappy compressed data'
- Fix float inconsistency with GP
- Use bankers rounding
- Make TeaContext safer
- Make logger thread-safe
- Build partial RowGroupFilter for and/or
- Set default
grpc_max_message_sizeto 16MB
- Support
byteaandbytea[]