PoC for understanding uniqueness constraints across partitions
Description
For ci_builds
we have a 1 unique index that prevent inserting jobs with duplicate tokens:
"index_ci_builds_on_token_encrypted" UNIQUE, btree (token_encrypted) WHERE token_encrypted IS NOT NULL
We need to understand how these indexes would work with multiple partitions because we use the token to uniquely identify a job.
Unique constraints identified
table | unique index | need to be fix for partitioning |
---|---|---|
ci_builds |
index_ci_builds_on_token_encrypted |
|
ci_builds_metadata |
index_ci_builds_metadata_on_build_id |
|
ci_job_artifacts |
index_ci_job_artifacts_on_job_id_and_file_type |
|
ci_pipeline_variables |
index_ci_pipeline_variables_on_pipeline_id_and_key |
|
ci_stages |
index_ci_stages_on_pipeline_id_and_name |
Objectives
Check if inserting resources scans all partitions and how it's affecting the performance.
Edited by Max Orefice