pqstream

pqstream is a program that streams changes out of a postgres database with the intent of populating other systems and enabling stream processing of data sets.

installation

$ go get -u github.com/tmc/pqstream/cmd/{pqs,pqsd}

basic usage

create an example database:

$ createdb dbname
# echo "create table notes (id serial, created_at timestamp, note text)" | psql dbname

connect the agent:

$ pqsd -connect postgresql://user:[email protected]/dbname

connect the cli:

$ pqs

at this point you will see streams of database operations rendered to stdout:

(in a psql shell):

dbname=# insert into notes values (default, default, 'here is a sample note');
INSERT 0 1
dbname=# insert into notes values (default, default, 'here is a sample note');
INSERT 0 1
dbname=# update notes set note = 'here is an updated note' where id=1;
UPDATE 1
dbname=# delete from notes where id = 1;
DELETE 1
dbname=#

our client should now show our operations:

$ pqs
{"schema":"public","table":"notes","op":"INSERT","id":"1","payload":{"created_at":null,"id":1,"note":"here is a sample note"}}
{"schema":"public","table":"notes","op":"INSERT","id":"2","payload":{"created_at":null,"id":2,"note":"here is a sample note"}}
{"schema":"public","table":"notes","op":"UPDATE","id":"1","payload":{"created_at":null,"id":1,"note":"here is an updated note"},"changes":{"note":"here is a sample note"}}
{"schema":"public","table":"notes","op":"DELETE","id":"1","payload":{"created_at":null,"id":1,"note":"here is an updated note"}}

field redaction

If there's a need to prevent sensitive fields (i.e. PII) from being exported the redactions flag can be used with pqsd:

$ pqsd -connect postgresql://user:[email protected]/dbname -redactions='{"public":{"users":["first_name","last_name","email"]}}'

The redactions is encoded in JSON and conforms to the following layout:

'{"schema":{"table":["field1","field2"]}}'`