Module system_catalog_dumper

Dumps catalog information to files. We run queries in serial rather than parallel because all queries in the pgwire connection are run in serial anyways. Running the queries in serial also makes cleaning up / aborting queries much easier.

Structs

• ClusterReplica
• Relation
• SqlPortForwardingInfo
• SystemCatalogDumper

Enums

• RelationCategory

Statics

• MAX_CLUSTER_REPLICA_ERROR_COUNT 🔒
  The maximum number of errors we tolerate for a cluster replica. If a cluster replica has more than this many errors, we skip it.
• PG_CONNECTION_TIMEOUT 🔒
• PG_QUERY_TIMEOUT 🔒
  Timeout for a query.
• RELATIONS 🔒
  This list is used to determine which relations to dump. The relations are grouped and delimited by their category (i.e. Basic object information)
• SELECT_CLUSTER_REPLICAS_QUERY 🔒
• SET_SEARCH_PATH_QUERY 🔒

Functions

• copy_relation_to_csv
• create_postgres_connection
• create_postgres_connection_string
  Create a postgres connection string. The following defaults are used if the arguments are not provided:
• format_catalog_dump_error_message 🔒
• format_file_path 🔒
• get_sql_port_forwarding_info
• query_column_names
• query_relation
• spawn_sql_port_forwarding_process
  Spawns a port forwarding process for the given k8s service. The process will retry if the port-forwarding fails and will terminate once the port forwarding reaches the max number of retries. We retry since kubectl port-forward is flaky.
• write_copy_stream