Skip to content

/ annotated.conf

Permalink
master
Switch branches/tags

annotated.conf/annotated.csv

Go to file
Upload for GUCS tutorial at SCALE 16x
Latest commit f845105 Mar 8, 2018 History
1 contributor

Users who have contributed to this file

270 lines (270 sloc) 70.6 KB
Raw Blame
category subcategory name type defaults recommendation range context docs comments
File Locations config_file string ConfigDir/postgresql.conf default postmaster Sets the server's main configuration file. Can only be changed via command-line switch for obvious reasons. Useful primarily for testing different configuration options, or for automated restart with different configuration options.
File Locations data_directory string ConfigDir default postmaster Sets the server's data directory. Supports the ability to distribute files according to sysadmin or operating system defined schemes, or for launching multiple restart instances using the same binaries. Most of the time, it's better to use configuration options to define these locations so that all PostgreSQL binaries default to the correct paths.
File Locations external_pid_file string None default postmaster Writes the postmaster PID to the specified file. Creates an extra copy of the process ID. Used for server administration tools which need a copy of the process ID in a specific directory.
File Locations hba_file string ConfigDir/pg_hba.conf default postmaster Sets the server's "hba" configuration file. Allows you to move the pg_hba file to a sysadmin-specified location.
File Locations ident_file string ConfigDir/pg_ident.conf default postmaster Sets the server's "ident" configuration file. Allows you to move the pg_ident file to a sysadmin-specified location.
Connections and Authentication Security and Authentication db_user_namespace bool off do not use sighup Enables per-database user names. This setting is a hack to work around the lack of per-database users in PostgreSQL. Unless you desperately need it, avoid this setting as it will eventually be replaced by something more maintainable.
Connections and Authentication Security and Authentication krb_caseins_users bool off varies sighup Sets whether Kerberos and GSSAPI user names should be treated as case-insensitive. Speak with your sysadmin or network security about how to set the various kerberos settings to match your local kerberos setup. Kerberos support must be compiled in to PostgreSQL, and set in pg_hba.conf.
Connections and Authentication Security and Authentication krb_server_keyfile string varies sighup Sets the location of the Kerberos server key file.
Connections and Authentication Security and Authentication password_encryption bool on on md5,scram-sha-256 user Encrypt passwords. There is no good reason for this to be set to “off”.
Connections and Authentication Security and Authentication ssl bool off on sighup Enables SSL connections. One of several different settings to turn on SSL connections for PostgreSQL. SSL is a very good idea for highly secure setups. In addition, you must compile in SSL support and set SSL connections in pg_hba.conf, as well as configuring SSL itself.
Connections and Authentication Security and Authentication ssl_ca_file config sighup Location of the SSL certificate authority file. You should always use SSL connections if you can. However, this does require setting up SSL.
Connections and Authentication Security and Authentication ssl_cert_file server.crt config sighup Location of the SSL server certificate file. According to your SSL configuration, which maybe provided by your installer.
Connections and Authentication Security and Authentication ssl_ciphers string HIGH:MEDIUM:+3DES:!aNULL default sighup Sets the list of allowed SSL ciphers. Allows DBAs to require “strong enough” or preset ciphers for SSL connections. If you have not compiled SSL support, this parameter will not be available.
Connections and Authentication Security and Authentication ssl_crl_file config sighup Location of the SSL certificate revocation list file. According to your SSL configuration, which maybe provided by your installer.
Connections and Authentication Security and Authentication ssl_dh_params_file config sighup Location of the SSL DH parameters file. According to your SSL configuration, which maybe provided by your installer.
Connections and Authentication Security and Authentication ssl_ecdh_curve prime256v1 config sighup Sets the curve to use for ECDH. According to your SSL configuration, which maybe provided by your installer.
Connections and Authentication Security and Authentication ssl_key_file server.key config sighup Location of the SSL server private key file. According to your SSL configuration, which maybe provided by your installer.
Connections and Authentication Security and Authentication ssl_prefer_server_ciphers on config sighup Give priority to server ciphersuite order. According to your SSL configuration, which maybe provided by your installer.
Connections and Authentication Connection Settings bonjour bool off off postmaster Enables advertising the server via Bonjour. Set to “on” if you've compiled in Bonjour support and have an application which works by autodiscovery of Postgres. Otherwise, leave off.
Connections and Authentication Connection Settings bonjour_name string machine name default postmaster Sets the Bonjour service name. Bonjour support must be compiled in and activated on the host machine to be live. You'll want alternate names if you have several instances of PostgreSQL on the same machine.
Connections and Authentication Connection Settings listen_addresses list localhost localhost,1 address postmaster Sets the host name or IP address(es) to listen to. Set your listen_address as restrictively as possible; '*' should only be used for development machines
Connections and Authentication Connection Settings max_connections integer 100 under 1000 1 to 262143 postmaster Sets the maximum number of concurrent connections. Should be set to the maximum number of connections which you expect to need at peak load. Note that each connection uses shared_buffer memory, as well as additional non-shared memory, so be careful not to run the system out of memory. In general, if you need more than 200 connections, you should probably be making more use of connection pooling.
Connections and Authentication Connection Settings port integer 5432 default 1 to 65535 postmaster Sets the TCP port the server listens on. Alternate ports are primarily useful for running several versions, or instances, of PostgreSQL on one machine. However, if you're using an alternate port to support several versions, it's often better to compile in the port number.
Connections and Authentication Connection Settings superuser_reserved_connections integer 3 default 0 to 262143 postmaster Sets the number of connection slots reserved for superusers. You should have at least one superuser connection open for troubleshooting at all times. So if you run more than two concurrent regular administrative tasks, you'll need more reserved connections. Note that this number is taken from max_connections, not in addition to it.
Connections and Authentication Connection Settings unix_socket_directories string /tmp change postmaster Sets the directories where Unix-domain sockets will be created. Change to a more secure directory, which many installers do for you.
Connections and Authentication Connection Settings unix_socket_group string NULL postgres postmaster Sets the owning group of the Unix-domain socket.
Connections and Authentication Connection Settings unix_socket_permissions integer 777 770 0 to 511 postmaster Sets the access permissions of the Unix-domain socket.
Connections and Authentication Connection Persistence authentication_timeout time 60 decrease 1s to 600s sighup Sets the maximum allowed time to complete client authentication. For production databases, it's important that this value be synchronized with the timeout on the application server side. Most web applications will want a shorter timeout, like 20s.
Connections and Authentication Connection Persistence tcp_keepalives_count integer 0 default 0 to 2147483647 user Maximum number of TCP keepalive retransmits. The three tcp_keepalive settings help manage a system which tends to have "undead" connection/query processes. For systems which support them, you can regulate checking that connections are still "live" end-to-end to kill them off. Not needed if you're not having a problem. Should be synchronized with the new TCP keepalive support in libpq on the client side.
Connections and Authentication Connection Persistence tcp_keepalives_idle integer 0 default 0s to 2147483647s user Time between issuing TCP keepalives.
Connections and Authentication Connection Persistence tcp_keepalives_interval integer 0 default 0s to 2147483647s user Time between TCP keepalive retransmits.
Resource Usage CPU max_parallel_workers integer 8 cores/2 0 to 1024 user Sets the maximum number of parallel workers than can be active at one time. … if you think you can benefit from parallel query, and even cores/1 for DW systems.
Resource Usage CPU max_worker_processes integer 8 increase 0 to 262143 postmaster Maximum number of concurrent worker processes. Increase to max_parallel_workers + other workers, such as workers for logical replication and custom background workers. Not more than your number of cores, though.
Resource Usage Memory dynamic_shared_memory_type ENUM auto-detect default posix,sysv,mmap,none postmaster Selects the dynamic shared memory implementation used.
Resource Usage Memory huge_pages ENUM try default off,on,try postmaster Use of huge pages on Linux. However, for small systems (< 2GB of RAM) may be beneficial to set to “off”.
Resource Usage Memory max_prepared_transactions integer 5 0 or max_connections 0 to 262143 postmaster Sets the maximum number of simultaneously prepared transactions. Most applications do not use XA prepared transactions, so should set this parameter to 0. If you do require prepared transactions, you should set this equal to max_connections to avoid blocking. May require increasing kernel memory parameters.
Resource Usage Memory shared_buffers memory varies 512kB to 8MB AvRAM / 4 168kB to 10737418238kB postmaster Sets the number of shared memory buffers used by the server. A memory quantity defining PostgreSQL's "dedicated" RAM, which is used for connection control, active operations, and more. However, since PostgreSQL also needs free RAM for file system buffers, sorts and maintenance operations, it is not advisable to set shared_buffers to a majority of RAM. Note that increasing shared_buffers often requires you to increase some system kernel parameters, most notably SHMMAX and SHMALL. See Operating System Environment: Managing Kernel Resources in the PostgreSQL documentation for more details. Also note that shared_buffers over 2GB is only supported on 64-bit systems.
Resource Usage Memory temp_buffers memory 8MB default 1008kB to 10737418238kB user Sets the maximum number of temporary buffers used by each session. Currently used only for holding temporary tables in memory. If your application requires heavy use of temporary tables (many proprietary reporting engines do) then you might want to increase this substantially. However, be careful because this is non-shared RAM which is allocated per session. Otherwise, the default is fine.
Resource Usage Memory work_mem integer 1MB ( AvRAM / max_connections ) OR ( AvRAM / 2 * max_connections ) 64kB to 2147483647kB user Sets the maximum memory to be used for query workspaces. Sets the limit for the amount of non-shared RAM available for each query operation, including sorts and hashes. This limit acts as a primitive resource control, preventing the server from going into swap due to overallocation. Note that this is non-shared RAM per operation, which means large complex queries can use multple times this amount. Also, work_mem is allocated by powers of two, so round to the nearest binary step. The second formula is for reporting and DW servers which run a lot of complex queries.
Resource Usage Kernel Resources max_files_per_process integer 1000 default 25 to 2147483647 postmaster Sets the maximum number of simultaneously open files for each server process. If you have a large database with many partitioned tables, you may want to increase this. Note that you will probably have to increase ulimits for the postgres user or system as well.
Resource Usage Kernel Resources max_stack_depth memory 2MB default 100kB to 2147483647kB superuser Sets the maximum stack depth, in kilobytes. Increase this if you have experienced the relevant error.
Resource Usage Kernel Resources shared_preload_libraries string default postmaster Lists shared libraries to preload into server. Primarily used for custom C libraries (data types, stored procedures) which you expect your application to use heavily. Trades memory overhead for these libraries against load time, so really should only be used for libraries you expect most queries to require.
Resource Usage Disk temp_file_limit memory -1 (disabled) 10GB -1kB to 2147483647kB superuser Limits the total size of all temporary files used by each process. … or something which is bigger than your largest possible sort, but not big enough to run you out of disk space.
WAL and Checkpoints Settings max_wal_size memory 1GB default 2MB to 2147483647MB sighup Sets the WAL size that triggers a checkpoint. … except for databases that write more than 1GB/hour of data, in which case increase the size of the log so that it's at least an hour worth of logs
WAL and Checkpoints Settings min_wal_size memory 80MB default 2MB to 2147483647MB sighup Sets the minimum size to shrink the WAL to.
WAL and Checkpoints Settings wal_compression bool off on superuser Compresses full-page writes written in WAL file. … unless your storage is less constrained than your CPU.
WAL and Checkpoints Settings wal_log_hints off off postmaster Writes full pages to WAL when first modified after a checkpoint, even for a non-critical modifications.
WAL and Checkpoints Archiving archive_command string (disabled) varies sighup Sets the shell command that will be called to archive a WAL file. All of the Archiving settings are part of a Point In Time Recovery or Warm Standby configuration. Please see the Backup and Restore section for more information.
WAL and Checkpoints Archiving archive_mode bool off varies always,on,off postmaster Allows archiving of WAL files using archive_command. Requires a restart to change, so if you want to turn archiving on and off, set this to 'on' and change archive_command instead. Even better, set archive_command to a script which can be disabled by trigger or ENV variable.
WAL and Checkpoints Archiving archive_timeout integer 0 varies 0s to 1073741823s sighup Forces a switch to the next WAL file if a new file has not been started within N seconds. Dependant on your tradeoff between disk space and letting the standby get behind.
WAL and Checkpoints Memory wal_buffers memory -1 auto default -18kB to 2621438kB postmaster Sets the number of disk-page buffers in shared memory for WAL. On very busy, high-core machines it can be useful to raise this to as much as 128MB.
WAL and Checkpoints Synch to Disk backend_flush_after memory 0 default 08kB to 2568kB user Number of pages after which previously performed writes are flushed to disk. Unless you have time to tune memory flushing behavior and test for improvements/regressions
WAL and Checkpoints Synch to Disk fsync bool on on sighup Forces synchronization of updates to disk. Never turn off unless your data is entirely disposable. Setting fsync=off is the equivalent of saying “I don't care about my data, I can recreate the database from scratch if I have to.” If synch activity is a performance concern, see synchronous_commit.
WAL and Checkpoints Synch to Disk full_page_writes bool on on sighup Writes full pages to WAL when first modified after a checkpoint. This is PostgreSQL's triple-check on transaction log integrity. Leave it on unless you have enough in-depth knowledge of your filesystem and hardware to be certain that torn page writes of log segments are completely prevented. Solaris/ZFS users claim to be able to turn this off, but that has not been destruction-tested.
WAL and Checkpoints Synch to Disk synchronous_commit bool on on local,remote_write,remote_apply,on,off user Sets the current transaction's synchronization level. If data integrity is less important to you than response times (for example, if you are running a social networking application or processing logs) you can turn this off, making your transaction logs asynchronous. This can result in up to wal_buffers or wal_writer_delay * 2 worth of data in an unexpected shutdown, but your database will not be corrupted. Note that you can also set this on a per-session basis, allowing you to mix “lossy” and “safe” transactions, which is a better approach for most applications.
WAL and Checkpoints Synch to Disk wal_sync_method string OS-dependent default fsync,fdatasync,open_sync,open_datasync sighup Selects the method used for forcing WAL updates to disk. On install, PostgreSQL figures out the best method for your OS. It's pretty good at this point; don't change the default. Note that the value of "fsync" shown in your postgresql.conf file is not necessarily the setting the server is using; try SHOW instead.
WAL and Checkpoints Synch to Disk wal_writer_delay integer 200 default 1ms to 10000ms sighup Time between WAL flushes performed in the WAL writer. Defines the maximum data (in time) that can be lost if synchronous_commit=off and the database shuts down. Because of long transactions, actual data lost can be up to twice this time. Has no effect if synchronous_commit=on. If you are going to turn synchronous_commit=off server-wide, you should probably lower this to prevent too much data loss.
WAL and Checkpoints Synch to Disk wal_writer_flush_after memory 1MB default 08kB to 21474836478kB sighup Amount of WAL written out by WAL writer that triggers a flush.
WAL and Checkpoints Commit Settings commit_delay integer 0 0 0 to 100000 superuser Sets the delay in microseconds between transaction commit and flushing WAL to disk. A primitive form of group commit without asynchronicity. Performance testing of this is very mixed; only set to non-zero if you have time to test the specific performance impact on your workload. Reasonable values are 200 to 1000.
WAL and Checkpoints Commit Settings commit_siblings integer 5 1 0 to 1000 user Sets the minimum concurrent open transactions before performing commit_delay. See commit_delay. Reasonable values are 3 to 8
WAL and Checkpoints Checkpoints checkpoint_completion_target real 1 default 0 to 1 sighup Time spent flushing dirty buffers during checkpoint, as fraction of checkpoint interval. Defines the fraction of one checkpoint_interval over which to spread checkpoints. The default value works for most users.
WAL and Checkpoints Checkpoints checkpoint_flush_after memory 256kB default 08kB to 2568kB sighup Number of pages after which previously performed writes are flushed to disk. Unless you have time to tune memory flushing behavior and test for improvements/regressions
WAL and Checkpoints Checkpoints checkpoint_timeout integer 300 default 30s to 86400s sighup Sets the maximum time between automatic WAL checkpoints. If you do really large ETL batches, you may want to increase this setting to the maximum length of a batch run.
WAL and Checkpoints Checkpoints checkpoint_warning integer 30 default 0s to 2147483647s sighup Enables warnings if checkpoint segments are filled more frequently than this.
WAL and Checkpoints Background Writer bgwriter_delay time 200 default 10ms to 10000ms sighup Background writer sleep time between rounds. Thanks to bgwriter autotuning, it should no longer be necessary for most users to touch the bgwriter settings. Only modify these if you have a demonstrated issue shown by checkpoint spikes and monitoring pg_stat_bgwriter. Laptop PostgreSQL users may want to increase bgwriter_delay to 60s to decrease I/O activity, since it is no longer possible to turn the bgwriter off.
WAL and Checkpoints Background Writer bgwriter_flush_after memory 512kB default 08kB to 2568kB sighup Number of pages after which previously performed writes are flushed to disk. Unless you have time to tune memory flushing behavior and test for improvements/regressions
WAL and Checkpoints Background Writer bgwriter_lru_maxpages integer 100 default 0 to 1073741823 sighup Background writer maximum number of LRU pages to flush per round.
WAL and Checkpoints Background Writer bgwriter_lru_multiplier real 2 default 0 to 10 sighup Multiple of the average buffer usage to free per round.
Replication Master Options max_replication_slots integer 10 max replicas * 2 0 to 262143 postmaster Sets the maximum number of simultaneously defined replication slots. Set to twice as many replicas as you ever expect to have.
Replication Master Options max_wal_senders integer 10 Max replicas you expect to have, doubled 0 to 262143 postmaster Sets the maximum number of simultaneously running WAL sender processes. If you are replicating, you want to set this to the maximum number of standby servers you might possibly have. Performance impact when set above zero, but no additional penalty for setting it higher.
Replication Master Options synchronous_standby_names string NULL default sighup Number of synchronous standbys and list of names of potential synchronous ones. Special, see syntax for sync standby config. Don't get into this if you're not sure what you're doing.
Replication Master Options track_commit_timestamp bool off on postmaster Collects transaction commit time.
Replication Master Options vacuum_defer_cleanup_age integer 0 0 0 to 1000000 sighup Number of transactions by which VACUUM and HOT cleanup should be deferred, if any. No longer effective thanks to hot_standby_feedback.
Replication Master Options wal_keep_segments integer 0 8 to 128 0 to 2147483647 sighup Sets the number of WAL files held for standby servers. … if using replication. Minimum number of WAL log segments to keep in order to support re-synchronizing streaming standby servers which have fallen behind or for an initial sync, a good rule of thumb is 4, or however many segments you go through in 30s, whichever is higher. This is in addition to max_wal_size, so make sure you have enough disk space. Not required if you are archiving logs.
Replication Master Options wal_level enum replica replica or logical minimal,replica,logical postmaster Set the level of information written to the WAL. Level replica is required for binary replication, and level logical is required for logical replication. This is a setting because raising the level adds more writes to the WAL, so if you’re not doing replication or archiving at all, set it to minimal.
Replication Master Options wal_sender_timeout time 60sec default 0ms to 2147483647ms sighup Sets the maximum time to wait for WAL replication.
Replication Standby Options hot_standby bool on on postmaster Allows connections and queries during recovery. Set to “on”, unless you want to specifically prohibit people from running queries on a standby server.
Replication Standby Options hot_standby_feedback bool off on sighup Allows feedback from a hot standby to the primary that will avoid query conflicts. Helps avoid query cancel on the replicas in most cases. Turn it off for a replica which does long-running reports and is allowed to lag.
Replication Standby Options max_logical_replication_workers integer 4 default 0 to 262143 postmaster Maximum number of logical replication worker processes. … unless logical replication is falling behind and the replica isn't handling other traffic
Replication Standby Options max_standby_archive_delay integer 30000 -1ms to 2147483647ms sighup Sets the maximum delay before canceling queries when a hot standby server is processing archived WAL data. If you are replicating primarily for failover, set this to a very low value (like 0) in order to keep the standby as up to date as possible. If this standby is running queries as its primary role, set to the length of time of the longest-running query you want to allow.
Replication Standby Options max_standby_streaming_delay integer 30000 -1ms to 2147483647ms sighup Sets the maximum delay before canceling queries when a hot standby server is processing streamed WAL data. If you are replicating primarily for failover, set this to a very low value (like 0) in order to keep the standby as up to date as possible. If this standby is running queries as its primary role, set to the length of time of the longest-running query you want to allow.
Replication Standby Options max_sync_workers_per_subscription integer 2 default 0 to 262143 sighup Maximum number of table synchronization workers per subscription. Consider raising to cores/2 when initially synchronizing logical replication for a new replica.
Replication Standby Options wal_receiver_status_interval time 10sec default 0s to 2147483s sighup Sets the maximum interval between WAL receiver status reports to the primary.
Replication Standby Options wal_receiver_timeout time 60sec default 0ms to 2147483647ms sighup Sets the maximum wait time to receive data from the primary.
Replication Standby Options wal_retrieve_retry_interval time 5sec default 1ms to 2147483647ms sighup Sets the time to wait before retrying to retrieve WAL after a failed attempt.
Query Tuning Planner Cost Constants cpu_index_tuple_cost real 0.005 0.001 0 to 1.79769e+308 user Sets the planner's estimate of the cost of processing each index entry during an index scan. Decrease this slightly to make your database favor indexes slightly more.
Query Tuning Planner Cost Constants cpu_operator_cost real 0.0025 0.0005 0 to 1.79769e+308 user Sets the planner's estimate of the cost of processing each operator or function call. Decrease this slightly to make your database favor indexes slightly more.
Query Tuning Planner Cost Constants cpu_tuple_cost real 0.01 default 0 to 1.79769e+308 user Sets the planner's estimate of the cost of processing each tuple (row).
Query Tuning Planner Cost Constants effective_cache_size integer 65536 ( AvRAM * 0.75 ) 18kB to 21474836478kB user Sets the planner's assumption about the size of the disk cache. Tells the PostgreSQL query planner how much RAM is estimated to be available for caching data, in both shared_buffers and in the filesystem cache. This setting just helps the planner make good cost estimates; it does not actually allocate the memory.
Query Tuning Planner Cost Constants random_page_cost real 4 default 0 to 1.79769e+308 user Sets the planner's estimate of the cost of a nonsequentially fetched disk page. Sets the ratio of seek to scan time for your database storage. Should not be altered unless you're using special storage (SSDs, high end SANs, etc.) where seek/scan ratios are actually different. If you need the database to favor indexes more, tune effective_cache_size and some of the cpu_* costs instead.
Query Tuning Planner Cost Constants seq_page_cost real 1 default 0 to 1.79769e+308 user Sets the planner's estimate of the cost of a sequentially fetched disk page. The main reason to modify seq_page_cost is to try to get planner costs to more-or-less indicate execution times in milleseconds. All other costs change relative to this cost automatically.
Query Tuning Planner Method Configuration enable_bitmapscan bool on default user Enables the planner's use of bitmap-scan plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_gathermerge bool on default user Enables the planner's use of gather merge plans.
Query Tuning Planner Method Configuration enable_hashagg bool on default user Enables the planner's use of hashed aggregation plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_hashjoin bool on default user Enables the planner's use of hash join plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_indexonlyscan bool on default user Enables the planner's use of index-only-scan plans.
Query Tuning Planner Method Configuration enable_indexscan bool on default user Enables the planner's use of index-scan plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_material bool on default user Enables the planner's use of materialization. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_mergejoin bool on default user Enables the planner's use of merge join plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_nestloop bool on default user Enables the planner's use of nested-loop join plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_seqscan bool on default user Enables the planner's use of sequential-scan plans. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_sort bool on default user Enables the planner's use of explicit sort steps. For interactive session use only when troubleshooting queries.
Query Tuning Planner Method Configuration enable_tidscan bool on default user Enables the planner's use of TID scan plans. For interactive session use only when troubleshooting queries.
Query Tuning Genetic Query Optimizer geqo bool on default user Enables genetic query optimization.
Query Tuning Genetic Query Optimizer geqo_effort integer 5 default 1 to 10 user GEQO: effort is used to set the default for other GEQO parameters.
Query Tuning Genetic Query Optimizer geqo_generations integer 0 default 0 to 2147483647 user GEQO: number of iterations of the algorithm.
Query Tuning Genetic Query Optimizer geqo_pool_size integer 0 default 0 to 2147483647 user GEQO: number of individuals in the population.
Query Tuning Genetic Query Optimizer geqo_seed real 0 0 to 1 user GEQO: seed for random path selection. If you set this manually, you can force repeatable execution paths for GEQO queries.
Query Tuning Genetic Query Optimizer geqo_selection_bias real 2 default 1.5 to 2 user GEQO: selective pressure within the population.
Query Tuning Genetic Query Optimizer geqo_threshold integer 12 15 2 to 2147483647 user Sets the threshold of FROM items beyond which GEQO is used. With new, faster processors it's tempting to raise the geqo_threshold a little, such as to 16 or 18. Increasing more than that is unwise as query planning time goes up geometrically.
Query Tuning Parallel Query force_parallel_mode ENUM off default off,on,regress user Forces use of parallel query facilities.
Query Tuning Parallel Query max_parallel_workers_per_gather integer 2 increase 0 to 1024 user Sets the maximum number of parallel processes per executor node. Increase if you plan to use parallel query to 4 or 8, depending on cores/concurrent sessions.
Query Tuning Parallel Query min_parallel_index_scan_size memory 512kB default 08kB to 7158278828kB user Sets the minimum amount of index data for a parallel scan.
Query Tuning Parallel Query min_parallel_table_scan_size memory 8MB default 08kB to 7158278828kB user Sets the minimum amount of table data for a parallel scan. … , unless doing IoT or a read-only database. Raise to 100MB or so if your traffic on the database is very bursty, to prevent the WAL from shrinking too much.
Query Tuning Parallel Query parallel_setup_cost real 1000 default 0 to 1.79769e+308 user Sets the planner's estimate of the cost of starting up worker processes for parallel query.
Query Tuning Parallel Query parallel_tuple_cost real 0.1 default 0 to 1.79769e+308 user Sets the planner's estimate of the cost of passing each tuple (row) from worker to master backend.
Query Tuning Other Planner Options constraint_exclusion bool partiton default partition,on,off user Enables the planner to use constraints to optimize queries. Default of “partition” is fine for most users. Setting it to “on” can allow optimization of UNION queries as well, but deserves testing before production deployment.
Query Tuning Other Planner Options cursor_tuple_fraction real 0 0 to 1 user Sets the planner's estimate of the fraction of a cursor's rows that will be retrieved. Increase this to 0.9 if most of the time you're using cursors to step through all of the rows of a query result.
Query Tuning Other Planner Options default_statistics_target integer 100 varies 1 to 10000 user Sets the default statistics target. Most applications can use the default of 100. For very small/simple databases, decrease to 10 or 50. Data warehousing applications generally need to use 500 to 1000. Otherwise, increase statistics targets on a per-column basis.
Query Tuning Other Planner Options effective_io_concurrency integer 1 varies 0 to 1000 user Number of simultaneous requests that can be handled efficiently by the disk subsystem. Set to the number of disks in your RAID array or number of I/O channels. Available only for platforms with posix_fadvise support (i.e. Linux). Currently only affects the execution of parallel bitmapscan, but might affect other I/O operations in future versions.
Query Tuning Other Planner Options from_collapse_limit integer 8 default 1 to 2147483647 user Sets the FROM-list size beyond which subqueries are not collapsed. While it's probably true that newer CPUs could support higher collapse_limits, there's not much incremental benefit to just raising either collapse_limit to 10 or 11.
Query Tuning Other Planner Options join_collapse_limit integer 8 default 1 to 2147483647 user Sets the FROM-list size beyond which JOIN constructs are not flattened. If for some reason you wanted to explicitly declare the join order for all of your queries, you could set this to 1. That is not recommended, though.
Query Tuning Other Planner Options replacement_sort_tuples integer 150000 0 0 to 2147483647 user Sets the maximum number of tuples to be sorted using replacement selection. Disable, this setting will be removed from Postgres 11.
Statistics Monitoring log_executor_stats bool off default superuser Writes executor performance statistics to the server log. Used for profiling the query executor.
Statistics Monitoring log_parser_stats bool off default superuser Writes parser performance statistics to the server log. Used for profiling the query parser.
Statistics Monitoring log_planner_stats bool off default superuser Writes planner performance statistics to the server log. Used for profiling the query planner.
Statistics Monitoring log_statement_stats bool off default superuser Writes cumulative performance statistics to the server log. Used for full query path profiling. Exclusive of the other three options.
Statistics Query and Index Statistics Collector stats_temp_directory string pg_stat_tmp default sighup Writes temporary statistics files to the specified directory. Useful for extremely high-volume databases; the stats temp directory could be set to a RAMdisk or other high-speed resource (at the cost of potentially losing some stats) as this file gets updated hundreds of times per second.
Statistics Query and Index Statistics Collector track_activities bool on default superuser Collects information about executing commands.
Statistics Query and Index Statistics Collector track_activity_query_size integer 1024 default 100 to 102400 postmaster Sets the size reserved for pg_stat_activity.query, in bytes. Sets the truncation threshold of queries in pg_stat_activity (and pg_stat_statements). Increase it if you have really long queries which are being cut off, but there is significant extra memory usage for keeping longer queries.
Statistics Query and Index Statistics Collector track_counts bool on default superuser Collects statistics on database activity. Needed for autovacuum to work properly. Do not turn off.
Statistics Query and Index Statistics Collector track_functions enum None pl none,pl,all superuser Collects function-level statistics on database activity. Set it to 'pl' to collect stats on user-defined functions. Very useful for stored procedure performance profiling and troubleshooting.
Statistics Query and Index Statistics Collector track_io_timing bool off default superuser Collects timing statistics for database I/O activity. Turn it on if you're monitoring disk usage per request.
Statistics Query and Index Statistics Collector update_process_title bool on default superuser Updates the process title to show the active SQL command. Updates the process title on OSes which support this. Very useful for checking resource usage by currently running queries.
Maintenance Memory maintenance_work_mem integer 16MB ( AvRAM / 8 ) for most 1024kB to 2147483647kB user Sets the maximum memory to be used for maintenance operations. Sets the limit for the amount that autovacuum, manual vacuum, bulk index build and other maintenance routines are permitted to use. Setting it to a moderately high value will increase the efficiency of vacuum and other operations. Applications which perform large ETL operations may need to allocate up to 1/4 of RAM to support large bulk vacuums. Note that each autovacuum worker may use this much, so if using multiple autovacuum workers you may want to decrease this value so that they can't claim over 1/8 or 1/4 of available RAM.
Maintenance Manual Vacuum vacuum_cost_delay integer 0 0ms to 100ms user Vacuum cost delay in milliseconds. Most of the time, you will want manual vacuum to execute without vacuum_delay, especially if you're using it as part of ETL. If for some reason you can't use autovacuum on an OLTP database, however, you may want to increase this to 20ms to decrease the impact vacuum has on currently running queries. Will cause vacuum to take up to twice as long to complete.
Maintenance Manual Vacuum vacuum_cost_limit integer 200 default 1 to 10000 user Vacuum cost amount available before napping.
Maintenance Manual Vacuum vacuum_cost_page_dirty integer 20 default 0 to 10000 user Vacuum cost for a page dirtied by vacuum.
Maintenance Manual Vacuum vacuum_cost_page_hit integer 1 default 0 to 10000 user Vacuum cost for a page found in the buffer cache.
Maintenance Manual Vacuum vacuum_cost_page_miss integer 10 default 0 to 10000 user Vacuum cost for a page not found in the buffer cache.
Maintenance Manual Vacuum vacuum_freeze_min_age integer 50000000 lower, varies 0 to 1000000000 user Minimum age at which VACUUM should freeze a table row. Most users will want to decrease this so that rows which have been cold for a long time get frozen earlier, and avoid an autovacuum_freeze. The suggestion of 500000 is for a moderately busy database; do not set to less than a few hours worth of XIDs. Maximum setting is 1/2 of autovaccuum_max_freeze_age.
Maintenance Manual Vacuum vacuum_freeze_table_age integer 150000000 400000000 0 to 2000000000 user Age at which VACUUM should scan whole table to freeze tuples. Generally set to 80% of autovacuum_max_freeze age to preempt a full vacuum freeze. If you can schedule cron vacuums during application slow periods, it might be valuable to lower this value in order to encourage vacuum freezing of tables before they are triggered by autovacuum.
Maintenance Manual Vacuum vacuum_multixact_freeze_min_age integer 5000000 lower, varies 0 to 1000000000 user Minimum age at which VACUUM should freeze a MultiXactId in a table row. Like freeze_min_age, lower this to somewhere around an hour of XID burn. Try starting with 500000.
Maintenance Manual Vacuum vacuum_multixact_freeze_table_age integer 150000000 350000000 0 to 2000000000 user Multixact age at which VACUUM should scan whole table to freeze tuples. Set to 80% of autovaccum_multixact_freeze_max_age
Maintenance Autovacuum autovacuum bool on on sighup Starts the autovacuum subprocess. Starts the daemon which cleans up your tables and indexes, preventing bloat and poor response times. The only reason to set it to “off” is for databases which regularly do large batch operations like ETL. Note that you can adjust the frequency or stop autovacuum on individual tables by adding rows to the pg_autovacuum system table.
Maintenance Autovacuum autovacuum_analyze_scale_factor real 0.1 default 0 to 100 sighup Number of tuple inserts, updates, or deletes prior to analyze as a fraction of reltuples. This setting should be optimal for most databases. However, very large tables (1m rows or more) in which rows are added in a skewed fashion may need to be autoanalyzed at a lower percentage, such as 5% or even 1%.
Maintenance Autovacuum autovacuum_analyze_threshold integer 50 default 0 to 2147483647 sighup Minimum number of tuple inserts, updates, or deletes prior to analyze.
Maintenance Autovacuum autovacuum_freeze_max_age integer 200000000 500000000 100000 to 2000000000 postmaster Age at which to autovacuum a table to prevent transaction ID wraparound. Triggers autovacuum automatically if a table is about to suffer from XID rollover. The setting is very conservative, and should probably be increased to 500million, but not higher.
Maintenance Autovacuum autovacuum_max_workers integer 3 default 1 to 262143 postmaster Sets the maximum number of simultaneously running autovacuum worker processes. If you have an installation with many tables (100's to 1000's) or with some tables which autovacuum takes hours to process, you may want to add additional autovacuum workers so that multiple tables can be vacuumed at once. Be conservative, though, as each autovacuum worker will utilize a separate CPU core, memory and I/O.
Maintenance Autovacuum autovacuum_multixact_freeze_max_age integer 400000000 default 10000 to 2000000000 postmaster Multixact age at which to autovacuum a table to prevent multixact wraparound. Triggers autovacuum automatically when the oldest “multixact” (a kind of lock transaction) is more than this old. Do not raise past 1billion.
Maintenance Autovacuum autovacuum_naptime integer 60 default 1s to 2147483s sighup Time to sleep between autovacuum runs. Decrease this to 30s or 15s if you have a large number (100's) of tables, or if you otherwise see from pg_stat_user_tables that autovacuum is not keeping up.
Maintenance Autovacuum autovacuum_vacuum_cost_delay integer 20 default -1ms to 100ms sighup Vacuum cost delay in milliseconds, for autovacuum. If autovacuum is having too much of a performance impact on running queries, you might want to increase this setting to 50ms. However, this will also cause individual vacuum tasks to take longer.
Maintenance Autovacuum autovacuum_vacuum_cost_limit integer -1 default -1 to 10000 sighup Vacuum cost amount available before napping, for autovacuum.
Maintenance Autovacuum autovacuum_vacuum_scale_factor real 0.2 default 0 to 100 sighup Number of tuple updates or deletes prior to vacuum as a fraction of reltuples.
Maintenance Autovacuum autovacuum_vacuum_threshold integer 50 default 0 to 2147483647 sighup Minimum number of tuple updates or deletes prior to vacuum.
Maintenance Autovacuum autovacuum_work_mem memory -1 AvRAM / workers / 4 -1kB to 2147483647kB sighup Sets the maximum memory to be used by each autovacuum worker process. Set a limit on this which is based on the number of autovac workers you expect to have running.
Reporting and Logging Where to Log log_destination string csvlog varies sighup Sets the destination for server log output. Your choice of log destination depends on your system administration plans and the status of your server. “syslog” or “eventlog” (Windows) are good choices for most development servers, because they can support centralized log monitors. For development and testing, however, “csvlog” is probably the most useful, as it allows you to run queries against the log contents.
Reporting and Logging Where to Log log_directory string pg_log varies sighup Sets the destination directory for log files. If you are having PostgreSQL keep its own activity logs on a production server, it's probably a good idea to locate them on separate storage from the database and transaction log.
Reporting and Logging Where to Log log_file_mode integer 600 default 0 to 511 sighup Sets the file permissions for log files. … unless you need to share the log with Postgres' unix group, in which case set it to 660.
Reporting and Logging Where to Log log_filename string postgresql-%Y-%m-%d_%H%M%S.log varies sighup Sets the file name pattern for log files. If you want your logs to rotate automatically without needing a cron job to delete old logs, try naming them after the days of the week or the month so they overwrite automatically (i.e. 'postgresql-%a' or 'postgresql-%d'). This also helps with log analysis.
Reporting and Logging Where to Log logging_collector bool on on postmaster Start a subprocess to capture stderr output and/or csvlogs into log files. Only relevant for “csvlog” and “stderr”.
Reporting and Logging Where to Log log_rotation_age integer 1440 default 0min to 35791394min sighup Automatic log file rotation will occur after N minutes. 1d is generally good for production. Set to 1h to rotate logs hourly when doing performance analysis.
Reporting and Logging Where to Log log_rotation_size memory 10MB 1GB 0kB to 2097151kB sighup Automatic log file rotation will occur after N kilobytes. Default is quite small if you have any extra logging turned on at all. Increase to avoid the creation of additional log segments with hard-to-predict names.
Reporting and Logging Where to Log log_truncate_on_rotation bool off on sighup Truncate existing log files of same name during log rotation. Set to “on” for production with a reusable logfile name to limit log accumulation if you don't have a sysadmin script to do so.
Reporting and Logging Where to Log syslog_facility string LOCAL0 varies local0,local1,local2,local3,local4,local5,local6,local7 sighup Sets the syslog "facility" to be used when syslog enabled. Change the logserver facility if you are having a conflict with other applications.
Reporting and Logging Where to Log syslog_ident string postgres POSTGRES_$HOST sighup Sets the program name used to identify PostgreSQL messages in syslog. If using a centralized logserver or if you have multiple Postgres instances, you probably want to identify your postgresql instance by hostname.
Reporting and Logging When to Log client_min_messages ENUM notice default debug5,debug4,debug3,debug2,debug1,log,notice,warning,error user Sets the message levels that are sent to the client. Unless doing interactive debugging, then you want it set to DEBUG1-5. If you have a client application which is confused by some of PostgreSQL's WARNINGs then you may want to set this to ERROR.
Reporting and Logging When to Log log_autovacuum_min_duration integer -1 1min -1ms to 2147483647ms sighup Sets the minimum execution time above which autovacuum actions will be logged. Logs all autovacuum actions which take more than the specified time. Useful for figuring out if autovacuum is bogging down your system or blocking.
Reporting and Logging When to Log log_error_verbosity ENUM default default terse,default,verbose superuser Sets the verbosity of logged messages. Unless doing intensive debugging. Alternately, set to TERSE if managing log volume is becoming a problem.
Reporting and Logging When to Log log_min_duration_statement integer -1 1min -1ms to 2147483647ms superuser Sets the minimum execution time above which statements will be logged. Possibly the most generally useful log setting for troubleshooting performance, especially on a production server. Records only long-running queries for analysis; since these are often your "problem" queries, these are the most useful ones to know about. Used for pg_fouine.
Reporting and Logging When to Log log_min_error_statement ENUM error default debug5,debug4,debug3,debug2,debug1,info,notice,warning,error,log,fatal,panic superuser Causes all statements generating error at or above this level to be logged. Logs SQL statements which error. If you have an application which routinely generates errors and can't fix it, then raise the level to FATAL or PANIC.
Reporting and Logging When to Log log_min_messages ENUM notice default debug5,debug4,debug3,debug2,debug1,info,notice,warning,error,log,fatal,panic superuser Sets the message levels that are logged. Unless doing serious troubleshooting. If you want to output parses and plans, set to DEBUG1.
Reporting and Logging What to Log debug_pretty_print bool off on user Indents parse and plan tree displays. For debugging a testing machine. Do not set in production.
Reporting and Logging What to Log debug_print_parse bool off default user Logs each query's parse tree. For debugging a testing machine. Do not set in production.
Reporting and Logging What to Log debug_print_plan bool off default user Logs each query's execution plan. For debugging a testing machine. Do not set in production.
Reporting and Logging What to Log debug_print_rewritten bool off default user Logs each query's rewritten parse tree. For debugging a testing machine. Do not set in production.
Reporting and Logging What to Log log_checkpoints bool off varies sighup Logs each checkpoint. When doing performance analysis, it's often a good idea to turn on most of the logging options and log them to a CSVlog.
Reporting and Logging What to Log log_connections bool off varies superuser-backend Logs each successful connection. Useful for performance analysis.
Reporting and Logging What to Log log_disconnections bool off varies superuser-backend Logs end of a session, including duration. Useful for performance analysis.
Reporting and Logging What to Log log_duration bool on varies superuser Logs the duration of each completed SQL statement. Useful for performance analysis.
Reporting and Logging What to Log log_hostname bool off off sighup Logs the host name in the connection logs. As this setting requires resolution of each connecting hostname, it's pretty much always too expensive to have on, even when troubleshooting.
Reporting and Logging What to Log log_line_prefix string default sighup Controls information prefixed to each log line. Primarily useful for providing extra information when logging to syslog or eventlog. Try "%h:%d:%u:%c %t" for this.
Reporting and Logging What to Log log_lock_waits bool off varies superuser Logs long lock waits. Useful for performance analysis.
Reporting and Logging What to Log log_replication_commands bool off on superuser Logs each replication command. … assuming you're monitoring replication status, which you should.
Reporting and Logging What to Log log_statement ENUM none varies none,ddl,mod,all superuser Sets the type of statements logged. For exhaustive performance analysis on test systems, set to 'all'. Most production setups will just want to use 'ddl' to make sure to record database-altering actions, but very secure setups may want to use 'mod' or even 'all'. Can produce a lot of log volume.
Reporting and Logging What to Log log_temp_files memory -1 varies -1kB to 2147483647kB superuser Log the use of temporary files larger than this number of kilobytes. This logger is used for troubleshooting sorts and other activities which are spilling to disk. If you use it at all, it's probably good to set it a something low like 1kB so that you know each query that spilled to disk, since any disk spill at all causes a dramatic slowdown in the query. Can be used to see if you need more work_mem, temp_mem or maintenance_work_mem.
Reporting and Logging What to Log log_timezone string per ENV local timezone sighup Sets the time zone to use in log messages. To avoid confusion, it's often useful to log to the timezone where the DBA or sysadmin lives.
Reporting and Logging What to Log syslog_sequence_numbers bool on default sighup Add sequence number to syslog messages to avoid duplicate suppression.
Reporting and Logging What to Log syslog_split_messages bool on default sighup Split messages sent to syslog by lines and to fit into 1024 bytes.
Lock Management deadlock_timeout time 1sec default 1ms to 2147483647ms superuser Sets the time to wait on a lock before checking for deadlock. Default is fine, except when you are troubleshooting/monitoring locks. In that case, you may want to lower it to as little as 50ms.
Lock Management max_locks_per_transaction integer 64 default 10 to 2147483647 postmaster Sets the maximum number of locks per transaction. Some databases with very complex schema or with many long-running tranactions need a higher amount. This is rare though.
Lock Management max_pred_locks_per_page integer 2 default 0 to 2147483647 sighup Sets the maximum number of predicate-locked tuples per page.
Lock Management max_pred_locks_per_relation integer -2 (disabled) -2147483648 to 2147483647 sighup Sets the maximum number of predicate-locked pages and tuples per relation.
Lock Management max_pred_locks_per_transaction integer 64 default 10 to 2147483647 postmaster Sets the maximum number of predicate locks per transaction. Raise if you have a lot of tables and are seeing some transactions fail, but modestly as a larger transaction table is expensive.
Locale & Formatting Display DateStyle list ISO, MDY default user Sets the display format for date and time values. Should be set according to the format in which you expect to receive date information.
Locale & Formatting Display extra_float_digits integer 0 default -15 to 3 user Sets the number of digits displayed for floating-point values. Only significant for applications which do a lot of float calculations, like scientific databases.
Locale & Formatting Display IntervalStyle enum postgres default postgres,postgres_verbose,sql_standard,iso_8601 user Sets the display format for interval values. This is just in case your applications are expecting something specific in how INTERVAL strings are output.
Locale & Formatting Display TimeZone string per ENV default user Sets the time zone for displaying and interpreting time stamps. To avoid a lot of confusion, make sure this is set to your local timeszone. If the server covers multiple time zones, then this should be set on a ROLE or connection basis.
Locale & Formatting Display timezone_abbreviations string Default default user Selects a file of time zone abbreviations. See appendencies for alternatives.
Locale & Formatting Locale client_encoding string per ENV default user Sets the client's character set encoding. Should match server_encoding unless you have a really good reason why not.
Locale & Formatting Locale lc_collate string as compiled N/A internal Shows the collation order locale. Set at initdb time. Displayed for information only.
Locale & Formatting Locale lc_ctype string as compiled N/A internal Shows the character classification and case conversion locale. Set at initdb time. Displayed for information only.
Locale & Formatting Locale lc_messages string as compiled default superuser Sets the language in which messages are displayed.
Locale & Formatting Locale lc_monetary string as compiled default user Sets the locale for formatting monetary amounts.
Locale & Formatting Locale lc_numeric string as compiled default user Sets the locale for formatting numbers.
Locale & Formatting Locale lc_time string as compiled default user Sets the locale for formatting date and time values.
Locale & Formatting Locale server_encoding string per ENV N/A internal Sets the server (database) character set encoding. Set at initdb time. Displayed for information only.
Other Settings & Defaults Identification application_name string psql name user Sets the application name to be reported in statistics and logs. Set this to a reasonable default for most user sessions; if in the middle of working over your application to support application names, this might be “unknown”.
Other Settings & Defaults Identification cluster_name string NULL set postmaster Sets the name of the cluster, which is included in the process title. Should be “postgres-1” or something else identifiable as this specific postmaster.
Other Settings & Defaults Identification event_source string NULL set postmaster Sets the application name used to identify PostgreSQL messages in the event log. Should be “postgres-1” or something else identifiable as this specific postmaster.
Other Settings & Defaults Default Locations default_tablespace string default user Sets the default tablespace to create tables and indexes in. Change this if you want a different tablespace for user-created tables. Generally, better set on a ROLE or session basis.
Other Settings & Defaults Default Locations search_path list "$user",public varies user Sets the schema search order for names that are not schema-qualified. Most DBAs either use the default or set search_path on a ROLE or database object basis. The one reason to set it in postgresql.conf is if you are taking the security step of removing the special "public" schema in order to lock down your database.
Other Settings & Defaults Default Locations temp_tablespaces list default user Sets the tablespace(s) to use for temporary tables and sort files. For applications which create lots of temporary objects, this setting can be used to put the temp space on a faster/separate device, or even a ramdisk. Because it accepts a list, it can even be used to load balance temp object creation among several tablespaces.
Other Settings & Defaults Libraries dynamic_library_path string $libdir default superuser Sets the path for dynamically loadable modules. Primarily useful if you've written lots of custom C libraries for your installation and want to organize them into custom directories.
Other Settings & Defaults Libraries local_preload_libraries string default user Lists unprivileged shared libraries to preload into each backend. This is largely a convenience setting, automatically loading libraries listed without needing an explicit load command. Has no effect on performance.
Other Settings & Defaults Replication session_replication_role ENUM origin default origin,replica,local superuser Sets the session's behavior for triggers and rewrite rules. Only gets changed for databases which are taking part in a replication chain. In that case,"origin" servers fire replication (and other) triggers, and "replica" do not. Part of the generic replication hooks which are used by Slony and Bucardo.
Other Settings & Defaults Statement Behavior default_transaction_deferrable bool off default user Sets the default deferrable status of new transactions. If you use serializable transactions by default, it may be also useful to set this in order to decrease the overhead of long-running transactions.
Other Settings & Defaults Statement Behavior default_transaction_isolation ENUM read committed default serializable,repeatable read,read committed,read uncommitted user Sets the transaction isolation level of each new transaction. Relates to transaction_isolation. Better set on a session or transaction basis as transaction_isolation in order to support specific types of transaction conflict resolution.
Other Settings & Defaults Statement Behavior default_transaction_read_only bool off default user Sets the default read-only status of new transactions. This setting is mainly useful for preventing yourself from accidentally changing data. It is not really a security setting, as anyone can revoke it on their own session. Better set on a session or ROLE level. Will show up as TRUE if you are on a replication standby.
Other Settings & Defaults Statement Behavior statement_timeout time 0 varies 0ms to 2147483647ms user Sets the maximum allowed duration of any statement. Defaults to 0, meaning no timeout. For most web applications, it's a good idea to set a default timeout, such as 60s to prevent runaway queries from bogging the server. If set, though, you need to remember to set (at the ROLE or session level) a higher statement_timeout for expected long-running maintenance or batch operations.
Other Settings & Defaults Statement Behavior transaction_deferrable bool off user Whether to defer a read-only serializable transaction until it can be executed with no possible serialization failures.
Other Settings & Defaults Statement Behavior transaction_isolation string read committed user Sets the current transaction's isolation level. Set per session if you need, for example, SERIALIZABLE semantics to prevent data conflicts for multi-step transactions.
Other Settings & Defaults Statement Behavior transaction_read_only bool off user Sets the current transaction's read-only status. Sets the current transaction to read only. Useful as part of a SQL injection prevention program. Shows as TRUE on replication standbys.
Other Settings & Defaults Text Search default_text_search_config string per ENV default user Sets default text search configuration. Set to the most common language used by the users, so that they don't have to pass the language parameter when calling TSearch functions.
Other Settings & Defaults Text Search gin_fuzzy_search_limit integer 0 varies 0 to 2147483647 user Sets the maximum allowed result for exact search by GIN. If you're going to use GIN queries in a web application, it's generally useful to set a limit on how many rows can be returned from the index just for response times. However, the maximum number needs to depend on your application; what do users see as an acceptable expression of "many"?
Other Settings & Defaults XML xmlbinary ENUM base64 varies base64,hex user Sets how binary values are to be encoded in XML. Set to whatever your client application supports.
Other Settings & Defaults XML xmloption ENUM content default content,document user Sets whether XML data in implicit parsing and serialization operations is to be considered as documents or content fragments.
Other Settings & Defaults Other Defaults check_function_bodies bool on default user Check function bodies during CREATE FUNCTION. You only really want to turn this off to resolve circular dependancies, and that can be done on a per-session basis. In general, checking for syntax errors in PL/pgSQL functions is a very good idea.
Other Settings & Defaults Other Defaults exit_on_error bool off default user Terminate session on any error.
Other Settings & Defaults Other Defaults gin_pending_list_limit memory 4MB default 64kB to 2147483647kB user Sets the maximum size of the pending list for GIN index. Unless you have a lot of GIN indexed data and have time to test the performance of fastupdate. Even then, it's probably better to set it on individual indexes.
Other Settings & Defaults Other Defaults restart_after_crash bool off default sighup Reinitialize server after backend crash. … unless deliberately running postgres in “ephemeral mode”
Other Settings & Defaults Other Defaults row_security bool on on user Enable row security. … except when testing row security policies.
Other Settings & Defaults Other Defaults session_preload_libraries string NULL default superuser Lists shared libraries to preload into each backend. Special uses for debugging or for loading application-specific extensions.
Other Settings & Defaults Timeouts idle_in_transaction_session_timeout time off 1hr 0ms to 2147483647ms user Sets the maximum allowed duration of any idling transaction. Set to 1 hour maximum, or as low as 1 minute if you know your query load well. Idle transactions are bad news.
Other Settings & Defaults Timeouts lock_timeout time 0 off default 0ms to 2147483647ms user Sets the maximum allowed duration of any wait for a lock. … but consider setting this per application or per query for any explicit locking attempts.
Other Settings & Defaults Timeouts old_snapshot_threshold time -1 (disabled) 2hrs -1min to 86400min postmaster Time before a snapshot is too old to read pages changed after the snapshot was taken. … or the length of the longest transaction you expect to run + 1 hour.
Other Settings & Defaults Previous PostgreSQL Versions operator_precedence_warning bool off default user Emit a warning for constructs that changed meaning since PostgreSQL 9.4.
Other Settings & Defaults Previous PostgreSQL Versions quote_all_identifiers bool off default user When generating SQL fragments, quote all identifiers.
Preset Options block_size integer 8192 N/A 8192 to 8192 internal Shows the size of a disk block. Informational: lets you know of non-standard installation or compile options.
Preset Options data_checksums bool off on internal Shows whether data checksums are turned on for this cluster. This has to be set at initdb time, and does create a significant amount of extra I/O. However, it will save you from a corrupt database down the line, so if you're not performance-constrained, always use it.
Preset Options integer_datetimes bool off N/A internal Datetimes are integer based. Informational: lets you know of non-standard installation or compile options.
Preset Options max_function_args integer 100 N/A 100 to 100 internal Shows the maximum number of function arguments. Informational: lets you know of non-standard installation or compile options.
Preset Options max_identifier_length integer 63 N/A 63 to 63 internal Shows the maximum identifier length. Informational: lets you know of non-standard installation or compile options.
Preset Options max_index_keys integer 32 N/A 32 to 32 internal Shows the maximum number of index keys. Informational: lets you know of non-standard installation or compile options.
Preset Options segment_size integer 131072 1310728kB to 1310728kB internal Shows the number of pages per disk file. Informational: lets you know of non-standard installation or compile options.
Preset Options server_version string 08/03/00 N/A internal Shows the server version. Informational: lets you know of non-standard installation or compile options.
Preset Options server_version_num integer 80300 N/A 100003 to 100003 internal Shows the server version as an integer. Informational: lets you know of non-standard installation or compile options.
Preset Options wal_block_size integer 8192 8192 to 8192 internal Shows the block size in the write ahead log. Informational: lets you know of non-standard installation or compile options.
Preset Options wal_segment_size integer 2048 20488kB to 20488kB internal Shows the number of pages per write ahead log segment. Informational: lets you know of non-standard installation or compile options.
Version and Platform Compatibility Other Platforms and Clients transform_null_equals bool off off user Treats "expr=NULL" as "expr IS NULL". Provided for compatibility with Microsoft Access and similar broken applications which treat "= NULL" as the same as "IS NULL".
Version and Platform Compatibility Previous PostgreSQL Versions array_nulls bool on default user Enable input of NULL elements in arrays. Provided for compatibility with 7.4 behavior.
Version and Platform Compatibility Previous PostgreSQL Versions backslash_quote ENUM safe_encoding default safe_encoding,on,off user Sets whether "\'" is allowed in string literals. If you have cleaned up your application code, you can set this to 'off' to help lock down the database. Older PHP applications will require the insecure setting of 'on'.
Version and Platform Compatibility Previous PostgreSQL Versions bytea_output enum hex hes escape,hex user Sets the output format for bytea.
Version and Platform Compatibility Previous PostgreSQL Versions default_with_oids bool off default user Create new tables with OIDs by default. Provided for consistency with 7.3 behavior. Since this creates an OID for every row, can cause OID wraparound in large databases.
Version and Platform Compatibility Previous PostgreSQL Versions escape_string_warning bool on off user Warn about backslash escapes in ordinary string literals. Useful for providing warnings for interpreted-language applications which may be engaging in unsafe string escape behavior. Unless you are currently porting or upgrading such an application, though, these warnings are not useful and should be turned off.
Version and Platform Compatibility Previous PostgreSQL Versions lo_compat_privileges bool off superuser Enables backward compatibility mode for privilege checks on large objects.
Version and Platform Compatibility Previous PostgreSQL Versions standard_conforming_strings bool off on user Causes '...' strings to treat backslashes literally. If you can clean up your application code, this disables use of \ as an escape character except in escaped (E' ') strings. This is both safer, and less likely to result in unexpected output for things like Windows filepaths.
Version and Platform Compatibility Previous PostgreSQL Versions synchronize_seqscans bool on on user Enable synchronized sequential scans. This new peformance enhancment can also cause rows to be returned in an order other than physical storage order. For poorly-written older applications, this may break application code; turn it off to disable.
Developer Options allow_system_table_mods bool off default postmaster Allows modifications of the structure of system tables. Only available in single-user mode; this setting is for initdb and may be used in the future for upgrade-in-place.
Developer Options debug_assertions bool off default internal Shows whether the running server has assertion checks enabled. Used for debugging PostgreSQL code problems; not for production use. Requires compile options.
Developer Options ignore_checksum_failure bool off default superuser Continues processing after a checksum failure. For rescuing a corrupt DB
Developer Options ignore_system_indexes bool off default backend Disables reading from system indexes. Useful for salvaging data from a corrupted database.
Developer Options post_auth_delay integer 0 default 0s to 2147s backend Waits N seconds on connection startup after authentication. Primarily used for attaching debuggers to sessions.
Developer Options pre_auth_delay integer 0 default 0s to 60s sighup Waits N seconds on connection startup before authentication. Primarily used for attaching debuggers to sessions.
Developer Options trace_notify bool off default user Generates debugging output for LISTEN and NOTIFY. The various TRACE options are for debugging specific behaviors interactively. Many of them require compile-time options. trace_notice is for debugging listen/notice.
Developer Options trace_recovery_messages enum log debug5,debug4,debug3,debug2,debug1,log,notice,warning,error sighup Enables logging of recovery-related debugging information. For troubleshooting replication/PITR failures.
Developer Options trace_sort bool off default user Emit information about resource usage in sorting. For debugging sorts.
Developer Options wal_consistency_checking string empty debugging only superuser Sets the WAL resource managers for which WAL consistency checks are done.
Developer Options zero_damaged_pages bool off off superuser Continues processing past damaged page headers. Used for salvaging data from a known-bad database. You should always make a binary backup before using this option, and it should not be used while users are allowed to connect. After damaged pages are erased, other kinds of data intergrity errors may persist (like broken PKs and FKs). ZDP should generally be used to get your DB to a stage where the data can be dumped and loaded into a new database.