Config​

Connection settings​

listen_addresses -> multi str

Specifies the TCP/IP address(es) on which the server is to listen for connections from client applications. If the list is empty, the server does not listen on any IP interface at all.

listen_port -> int16

The TCP port the server listens on; 5656 by default. Note that the same port number is used for all IP addresses the server listens on.

Resource usage​

effective_io_concurrency -> int64

Sets the number of concurrent disk I/O operations that can be executed simultaneously. Corresponds to the PostgreSQL configuration parameter of the same name.

query_work_mem -> cfg::memory

The amount of memory used by internal query operations such as sorting. Corresponds to the PostgreSQL work_mem configuration parameter.

shared_buffers -> cfg::memory

The amount of memory the database uses for shared memory buffers. Corresponds to the PostgreSQL configuration parameter of the same name. Changing this value requires server restart.

Query planning​

default_statistics_target -> int64

Sets the default data statistics target for the planner. Corresponds to the PostgreSQL configuration parameter of the same name.

effective_cache_size -> cfg::memory

Sets the planner’s assumption about the effective size of the disk cache that is available to a single query. Corresponds to the PostgreSQL configuration parameter of the same name.

Query behavior​

allow_bare_ddl -> cfg::AllowBareDDL

Allows for running bare DDL outside a migration. Possible values are cfg::AllowBareDDL.AlwaysAllow and cfg::AllowBareDDL.NeverAllow.

When you create an instance, this is set to cfg::AllowBareDDL.AlwaysAllow until you run a migration. At that point it is set to cfg::AllowBareDDL.NeverAllow because it’s generally a bad idea to mix migrations with bare DDL.

apply_access_policies -> bool

Determines whether access policies should be applied when running queries. Setting this to false effectively puts you into super-user mode, ignoring any access policies that might otherwise limit you on the instance.

This setting can also be conveniently accessed via the “Config” dropdown menu at the top of the EdgeDB UI (accessible by running the CLI command edgedb ui from within a project). The setting will apply only to your UI session, so you won’t have to remember to re-enable it when you’re done.

force_database_error -> str

A hook to force all queries to produce an error. Defaults to ‘false’.

This parameter takes a str instead of a bool to allow more verbose messages when all queries are forced to fail. The database will attempt to deserialize this str into a JSON object that must include a type (which must be an EdgeDB error type name), and may also include message, hint, and details which can be set ad-hoc by the user.

For example, the following is valid input:

'{ "type": "QueryError", "message": "Did not work", "hint": "Try doing something else", "details": "Indeed, something went really wrong" }'

As is this:

'{ "type": "UnknownParameterError" }'

Client connections​

allow_user_specified_id -> bool

Makes it possible to set the .id property when inserting new objects.

Enabling this feature introduces some security vulnerabilities:

1. An unprivileged user can discover ids that already exist in the database by trying to insert new values and noting when there is a constraint violation on .id even if the user doesn’t have access to the relevant table.

2. It allows re-using object ids for a different object type, which the application might not expect.

Additionally, enabling can have serious performance implications as, on an insert, every object type must be checked for collisions.

As a result, we don’t recommend enabling this. If you need to preserve UUIDs from an external source on your objects, it’s best to create a new property to store these UUIDs. If you will need to filter on this external UUID property, you may add an index on it.

session_idle_timeout -> std::duration

Sets the timeout for how long client connections can stay inactive before being forcefully closed by the server.

Time spent on waiting for query results doesn’t count as idling. E.g. if the session idle timeout is set to 1 minute it would be OK to run a query that takes 2 minutes to compute; to limit the query execution time use the query_execution_timeout setting.

The default is 60 seconds. Setting it to <duration>'0' disables the mechanism. Setting the timeout to less than 2 seconds is not recommended.

Note that the actual time an idle connection can live can be up to two times longer than the specified timeout.

This is a system-level config setting.

session_idle_transaction_timeout -> std::duration

Sets the timeout for how long client connections can stay inactive while in a transaction.

The default is 10 seconds. Setting it to <duration>'0' disables the mechanism.

query_execution_timeout -> std::duration

Sets a time limit on how long a query can be run.

Setting it to <duration>'0' disables the mechanism. The timeout isn’t enabled by default.

db> 

select cfg::Config.apply_access_policies;

{true}

db> 

configure session set apply_access_policies := false;

OK: CONFIGURE SESSION

db> 

select cfg::Config.apply_access_policies;

{false}

db> 
... 

configure instance insert
  Auth {priority := 0, method := (insert Trust)};

OK: CONFIGURE INSTANCE

db> 
... 

configure instance insert
  Auth {priority := 0, method := (insert Trust)};

OK: CONFIGURE INSTANCE

db> 
... 

configure instance insert
  Auth {priority := 0, method := (insert SCRAM)};

OK: CONFIGURE INSTANCE

db> 

select <cfg::memory>'1B'; # 1 byte

{<cfg::memory>'1B'}

db> 

select <cfg::memory>'5KiB'; # 5 kibibytes

{<cfg::memory>'5KiB'}

db> 

select <cfg::memory>'128MiB'; # 128 mebibytes

{<cfg::memory>'128MiB'}