citusdata / pg_shard

96%

develop: 95%

LAST BUILD ON BRANCH feature-reject_zero_shard_insert

branch: feature-reject_zero_shard_insert

CHANGE BRANCH

x

Reset

• feature-reject_zero_shard_insert
• develop
• feature-drop_distributed_table-#78_C_UtilityHook
• master
• feature-composite_types_bugs#80
• bugfix-copy-script-#95
• feature-bidi_md_sync#27
• bugfix-deparse_constraint#94
• feature-check_name_length
• feature-check_partition_type
• feature-metadata_uses_spi
• sql_modification_poc
• feature-add_citusdb_tests
• increase_coverage
• push_simple_aggregates_down
• push_simple_aggregates_down_unified
• feature-remove_unsigned
• feature-remove_templates
• feature-reorg
• release-1.1.1
• v1.1.1
• use_regclass
• feature-support_plpgsql
• feature-turn_off_unused_warning
• feature-metadata_locks
• aggregate_changes
• release-1.2.0
• v1.2.0
• EXPERIMENTAL_push_simple_aggragates
• release-1.2.1
• v1.2.1
• release-1.2.2
• feature-default_port
• named_params
• feature-9.5_compatibility
• fix_includes
• better_hash_token_distribution
• warnings
• v1.2.2
• feature-drop-table
• feature-drop-table-partition-only
• feature-test_on_citusdb41
• transaction_manager_integration
• fix_hash_token_distribution
• redhat_regression_tests
• feature-cache_spi_plans
• insert_throughput_readme
• release-1.2.3
• feature-9.5_beta_fix
• feature-hidden_visibility
• v1.2.3
• feature-copy
• feature-95_final_compat
• feature-use_citus_tools
• documentation/fix_parallel_command
• deprecation
• ozgune-patch-1
• develop_candidate

Build # 389

Build Type

push

travis-ci

Committed by jasonmp85

Commit Message

Throw error for zero-shard INSERTs

Found this when doing CitusDB integration. If CitusDB has created an
e.g. range partition table I found it "just works" with pg_shard save
one caveat: if no shard exists to receive an incoming row, our zero-
shard SELECT logic kicked in and pg_shard would happily put the row
into the local relation.

UPDATE and DELETE are safe to run locally as the local table should be
empty to begin with, but INSERT is not: we should raise an error rather
than continue with a zero-shard INSERT.

Run Details

2061 of 2156 relevant lines covered (95.59%)

747.84 hits per line