Thread: BUG #1930: Cannot create foreign key constraint
The following bug has been logged online: Bug reference: 1930 Logged by: Warren Little Email address: warren.little@reardeninc.com PostgreSQL version: 8.0.3 Operating system: 2.6.10-1.770_FC3smp #1 SMP Thu Feb 24 18:36:43 EST 2005 x86_64 x86_64 x86_64 GNU/Linux Description: Cannot create foreign key constraint Details: pg_restore generates the following error: pg_restore: [archiver (db)] Error from TOC entry 2950; 16386 17827531 FK CONSTRA INT asientry_r0 tigris pg_restore: [archiver (db)] could not execute query: ERROR: insert or update on table "asientry" violates foreign key constraint "asientry_r0" DETAIL: Key (pid)=(AA0003403544) is not present in table "pobject". Command was: ALTER TABLE ONLY asientry ADD CONSTRAINT asientry_r0 FOREIGN KEY (pid) REFERENCES pobject(pid); Also running the alter table after restore generates same constraint violation error. The following query returns one row: select * from asientry, pobject where pobject.pid = 'AA0003403544' and pobject.pid = asientry.pid; Any ideas
"Warren Little" <warren.little@reardeninc.com> writes: > pg_restore: [archiver (db)] could not execute query: ERROR: insert or > update on table "asientry" violates foreign key constraint "asientry_r0" > DETAIL: Key (pid)=(AA0003403544) is not present in table "pobject". > Command was: ALTER TABLE ONLY asientry > ADD CONSTRAINT asientry_r0 FOREIGN KEY (pid) REFERENCES pobject(pid); > Also running the alter table after restore generates same constraint > violation error. > The following query returns one row: > select * from asientry, pobject > where pobject.pid = 'AA0003403544' > and pobject.pid = asientry.pid; That seems a bit odd. Are the two pid columns exactly the same datatype (and what is it anyway)? What plan does EXPLAIN show for that SELECT, and what plan do you see for SELECT fk.pid FROM ONLY asientry fk LEFT OUTER JOIN ONLY pobject pk ON pk.pid=fk.pid WHERE pk.pid IS NULL AND fk.pid IS NOT NULL; (which is what is probably being used to test foreign key validity)? regards, tom lane