Catherine Quamme

To get sent to Postgres, it would need to be base 64 encoded. You could use the following: java.util.Base64.getEncoder().encodeToString(zipbytea)

When a column is part of the primary key you will need to include both new and old value, otherwise you will get problems with conflict detection.

People using transforms often need to set the following parameter to false: dataloader.apply.changes.only Indicates that old data should be used to create the update statement. If old data is equal to the new data and this property is set to true, then no update statement will be run.

This situation is not currently handled, the parameter creates a SQL event that is run and doesn't check if the table exists or not.

Is node.offline true on both sides? This will need to be true on both sides. For example, if you have a server and a client-1 and client-1 you want to be offline, then you would insert into sym_parameter (node_group_id, external_id, param_key, param_value) values ('client', '1', 'node.offline', 'true'); You could also turn on debug logging to see what sql is being run. You turn this on in conf/log4j2.xml

Is the contents of SymmetricDS tables expected to be copied manually to the new offline node at least once? - Yes. The design was to be online when you register the node, if you want the node to always be offline then you would need to import config on it, and manually register the node and put in entries in sym_node, sym_node_security, and sym_node_identity.

If you check your log, do you see the Purge Service running to completion? What do you have the purge retention minutes set to?

You will either need to get rid of the 1st or 3rd trigger. If you only want to send an initial load and not capture changes after that, get rid of the 1st trigger. If you want to keep capturing changes, get rid of the 3rd trigger.