Skip to content

Spock Configuration

These parameters control how the Spock extension handles replication behavior in your pgEdge cluster.

exception_behaviour

  • Type: String
  • Default: transdiscard
  • Options: discard, transdiscard, sub_disable

  • Description: This parameter defines how Spock handles replication exceptions when they occur during data synchronization. The available options provide different levels of intervention:

  • discard skips only the offending statement and continues replication.

  • transdiscard skips the entire offending transaction and continues.
  • sub_disable disables the subscription and requires manual intervention to re-enable it.

See the Spock documentation for detailed information about exception handling behavior.

In the following example, the inventory sets the exception behavior to discard transactions:

pgedge:
  vars:
    exception_behaviour: transdiscard

pgedge_join_zones

  • Type: List of integers
  • Default: none
  • Description: This parameter names the zones to join to a cluster that already holds data, in the order they should join. Each zone must be in the inventory. The deployment copies every database in db_names into each zone in turn, while the existing zones keep taking writes, and then makes it a full member of the cluster. A listed zone that is already a member is reported and skipped. Pass it on the command line rather than in the inventory, and use it only on a cluster that already holds data: a run with it refuses to start when no database holds data yet. Without it, a run refuses a zone that is in the inventory but not yet part of a cluster that holds data. See Adding a Zone to a Running Cluster.

The following examples join zone 4, and then zones 4 and 5 in that order:

ansible-playbook -i inventory.yaml playbook.yaml -e pgedge_join_zones=4
ansible-playbook -i inventory.yaml playbook.yaml -e pgedge_join_zones=4,5
ansible-playbook -i inventory.yaml playbook.yaml -e '{"pgedge_join_zones": [4, 5]}'

A bracketed list given as a plain argument, such as -e pgedge_join_zones=[4], reaches Ansible as text, and the run refuses it.

pgedge_join_wait_minutes

  • Type: Integer
  • Default: 30
  • Description: This parameter sets how long a join waits, in minutes, for a zone to apply a bookmark another zone has set, and for each new subscription to start replicating. A join uses these waits to confirm that every write has arrived before it moves on, and they normally take seconds. A wait that runs out fails the join in a way that the same command, run again, resumes. The copy itself is governed by pgedge_seed_stall_minutes and pgedge_seed_max_hours instead.

pgedge_seed_zone

  • Type: Integer
  • Default: none
  • Description: This parameter is for recovery only. In an HA cluster, it names a zone that already holds the cluster's data while every other zone is empty and takes no writes. The empty zones copy the seed zone's schema and data before the rest of the subscriptions are created. The recovery playbook sets it to the zone it restores; an ordinary deployment leaves it unset. To add a zone to a cluster that is running, use pgedge_join_zones instead. A run given both refuses to start.

pgedge_seed_stall_minutes

  • Type: Integer
  • Default: 15
  • Description: This parameter sets how long a Spock copy may go without progress before the role gives up, in minutes. It applies to the copy from pgedge_seed_zone in a recovery and to the copy a zone named in pgedge_join_zones makes from an existing zone. Progress is any change in the sync step, the tables left to sync, the database's size, or the COPY and index-build progress Postgres reports. The role gives up sooner if Spock disables the subscription or its apply worker stays down for two minutes. The limit applies to each database in db_names separately. A join that gives up this way can be resumed by running the same command again, as long as Spock's copy is still going.

pgedge_seed_max_hours

  • Type: Integer
  • Default: 24
  • Description: This parameter limits how long a Spock copy may take, in hours, for each database in db_names separately: the copy from pgedge_seed_zone in a recovery, or a joining zone's copy. It is a backstop, not an estimate: the copy takes as long as the database is large.