问题
I am writing a Spring Boot web-app and using a Postgres db to persist my data. I created a table in Postgres using create table user (id bigserial primary key not null, name text not null;
and identified its sequence_name
by looking at the schema (in this case, it is user_id_seq
). Then, in my User
entity class in Spring Boot, I added the following:
@Entity
@Table(name = "user")
public class User implements Serializable {
@Id
@SequenceGenerator(name = "user_local_seq", sequenceName = "user_id_seq", allocationSize = 1)
@GeneratedValue(strategy = GenerationType.SEQUENCE, generator = "user_local_seq")
private Long id;
...
making sure that the sequenceName
matches what I saw earlier. Now when I start my spring boot app, I am able to successfully boot it but I get the following "error" in the trace:
main] org.hibernate.tool.hbm2ddl.SchemaExport : ERROR: sequence "user_id_seq" does not exist
I killed the app and started it again and this time, I got:
main] org.hibernate.tool.hbm2ddl.SchemaExport : HHH000389: Unsuccessful: drop sequence user_id_seq
main] org.hibernate.tool.hbm2ddl.SchemaExport : ERROR: sequence "user_id_seq" does not exist
What does this mean? Am I missing something? Any help/insight is appreciated.
回答1:
Here is insight.
ERROR: sequence "user_id_seq" does not exist
It mean your sequence either not exist in database OR the user doesn't has permission to access it.
Solution:
- Check
user_id_seq
in database by command\ds
- Grant access on sequence to specific user.
GRANT ALL ON ALL SEQUENCES IN SCHEMA schema_name TO user_name;
回答2:
Make sure your search_path is set to the schema that the sequence belongs to. Even if the table in question is schema-qualified, the search_path must also be set appropriately.
回答3:
Guessing at the backend that's abstracted away by spring boot, but I found a similar issue with the same error while using postgresql 9.6 that may be of use. When my table has been rebuilt (DROP and then CREATE with same sequence name) the has_sequence_privilege(name,text,text) can fail to recognize the pg_class.relname of the sequence.
Using the pg_class.oid resolved this for me.
To clarify:
SELECT has_sequence_privilege('my_role_name',c.relname,'USAGE')
FROM pg_namespace n
JOIN pg_class c
ON c.relnamespace = n.oid
WHERE c.relnamespace = n.oid AND c.relkind = 'S'::"char"
AND n.nspname = 'my_schema_name'
throws an error for rebuilt tables, while
SELECT has_sequence_privilege('my_role_name',c.oid,'USAGE')
FROM pg_namespace n
JOIN pg_class c
ON c.relnamespace = n.oid
WHERE c.relnamespace = n.oid AND c.relkind = 'S'::"char"
AND n.nspname = 'my_schema_name'
does not.
回答4:
Put database-specific escape-characters ( " in postgres) around the name of the sequence-name.
The name of the sequence has been changed by the driver. This sometimes happens because the Driver thinks the db-server is case-insensitive but the db-server is case-sensitive.
@Id
@GeneratedValue(strategy = GenerationType.SEQUENCE, generator = "TEACHER")
@SequenceGenerator(name = "TEACHER", sequenceName = "\"Teahcer_pkey\"")
@Column(name = "id", nullable = false)
来源:https://stackoverflow.com/questions/44962937/sequence-does-not-exist-when-it-does-postgres-spring-boot