keithf4
commented
4 years ago What version of pg_partman are you using? You can see this by running \dx in psql. Also with daily partitioning, maintenance should be running at least once, preferrably twice, a day to properly keep ahead, even with a 30 day premake.
Would you also mind providing the \d+ tablename output for the parent table so I can try to recreate things as you have them? Thank you.
keith@keith=# \dx
List of installed extensions
Name | Version | Schema | Description
---------------+---------+------------+--------------------------------------------------------------
dblink | 1.2 | public | connect to other PostgreSQL databases from within a database
pg_partman | 4.2.2 | partman | Extension to manage partitioned tables by time or ID
pgstattuple | 1.5 | public | show tuple-level statistics
pgtap | 1.1.0 | public | Unit testing for PostgreSQL
plpgsql | 1.0 | pg_catalog | PL/pgSQL procedural language
KevMurph
Hello, I have had a recurring problem with pg_partman maintenance running against PostgreSQL 10. I feel as though sometimes the issue crops up more often when I wait a week between running maintenance for the partitions, but that may be a coincidence.
Summary: Run maintenance fails with 'Unterminated Format Specifier'
Here are the run_maintenance logs and part_config setup:
Run maintenance error:
NOTICE: run_maint: parent_table: my_schema_name.my_table_name, v_last_partition: my_table_name_p2019_12_23 NOTICE: run_maint: v_current_partition_timestamp: 2019-12-12 19:08:52.48493+00, v_max_time_parent:
NOTICE: run_maint before loop: current_partition_timestamp: 2019-12-12 19:08:52.48493+00, v_premade_count: 10, v_sub_timestamp_min: , v_sub_timestamp_max:
NOTICE: run_maint: parent_table: my_schema_name.my_table_name, v_premade_count: 10, v_next_partition_timestamp: 2019-12-23 00:00:00+00
NOTICE: create_partition_time: v_partition_expression: activity_date
NOTICE: create_partition_time v_sql: CREATE TABLE my_schema_name.my_table_name_p2019_12_24 (LIKE my_schema_name.my_table_name INCLUDING DEFAULTS INCLUDING CONSTRAINTS INCLUDING STORAGE INCLUDING COMMENTS INCLUDING INDEXES)
NOTICE: ALTER TABLE my_schema_name.my_table_name_p2019_12_24 SET TABLESPACE pg_default;
ERROR: unterminated format() type specifier
CONTEXT: PL/pgSQL function inherit_template_properties(text,text,text) line 105 at assignment
SQL statement "SELECT partman.inherit_template_properties(p_parent_table, v_parent_schema, v_partition_name)"
PL/pgSQL function create_partition_time(text,timestamp with time zone[],boolean,boolean) line 216 at PERFORM
PL/pgSQL function partman.run_maintenance(text,boolean,boolean,boolean) line 241 at assignment
DETAIL:
HINT: For a single "%" use "%%".
CONTEXT: PL/pgSQL function create_partition_time(text,timestamp with time zone[],boolean,boolean) line 441 at RAISE
PL/pgSQL function partman.run_maintenance(text,boolean,boolean,boolean) line 241 at assignment
DETAIL:
HINT:
CONTEXT: PL/pgSQL function partman.run_maintenance(text,boolean,boolean,boolean) line 358 at RAISE
part_config:
-[ RECORD 1 ]--------------+--------------------------- parent_table | my_schema_name.my_table_name control | activity_date partition_type | native partition_interval | 1 day constraint_cols | premake | 30 optimize_trigger | 4 optimize_constraint | 30 epoch | none inherit_fk | t retention | 60 days retention_schema | retention_keep_table | t retention_keep_index | f infinite_time_partitions | t datetime_string | YYYY_MM_DD automatic_maintenance | on jobmon | t sub_partition_set_full | f undo_in_progress | f trigger_exception_handling | f upsert | trigger_return_null | t template_table | my_schema_name.template_my_table_name publications |
Has anyone seen this before? Any recommendations?