Faster logical replication tests.

Logical replication status can take wal_receiver_status_interval
seconds to get updated. Default is 10s, which means tests in
which logical replication is used can take a long time to finish.
We reduce it to 1 second to speed these tests up.

Logical replication apply launcher launches workers every
wal_retrieve_retry_interval, so if we have many shard moves with
logical replication consecutively, they will be throttled by this
parameter. Default is 5s, we reduce it to 1s so we finish tests
faster.

(cherry picked from commit 0e0fd6599a)
pull/4578/head
Hadi Moshayedi 2021-01-15 12:00:50 -08:00 committed by Sait Talha Nisanci
parent a603ad9cbf
commit 043c3356ae
1 changed files with 9 additions and 0 deletions

View File

@ -418,6 +418,15 @@ push(@pgOptions, "max_parallel_workers_per_gather=0");
# Allow CREATE SUBSCRIPTION to work # Allow CREATE SUBSCRIPTION to work
push(@pgOptions, "wal_level='logical'"); push(@pgOptions, "wal_level='logical'");
# Faster logical replication status update so tests with logical replication
# run faster
push(@pgOptions, "wal_receiver_status_interval=1");
# Faster logical replication apply worker launch so tests with logical
# replication run faster. This is used in ApplyLauncherMain in
# src/backend/replication/logical/launcher.c.
push(@pgOptions, "wal_retrieve_retry_interval=1000");
# Citus options set for the tests # Citus options set for the tests
push(@pgOptions, "citus.shard_count=4"); push(@pgOptions, "citus.shard_count=4");
push(@pgOptions, "citus.shard_max_size=1500kB"); push(@pgOptions, "citus.shard_max_size=1500kB");