]> granicus.if.org Git - postgresql/commit
Fix timeout handling in logical replication worker
authorMichael Paquier <michael@paquier.xyz>
Fri, 18 Oct 2019 05:26:53 +0000 (14:26 +0900)
committerMichael Paquier <michael@paquier.xyz>
Fri, 18 Oct 2019 05:26:53 +0000 (14:26 +0900)
commit04510dbe34b9d8eef1754413806c56fcc1d369ed
treef0d36f26d6d9decf1685d33c7d090922d8354f3a
parent1391c13ce3b898caad07261223bb75f8390c45e2
Fix timeout handling in logical replication worker

The timestamp tracking the last moment a message is received in a
logical replication worker was initialized in each loop checking if a
message was received or not, causing wal_receiver_timeout to be ignored
in basically any logical replication deployments.  This also broke the
ping sent to the server when reaching half of wal_receiver_timeout.

This simply moves the initialization of the timestamp out of the apply
loop to the beginning of LogicalRepApplyLoop().

Reported-by: Jehan-Guillaume De Rorthais
Author: Julien Rouhaud
Discussion: https://postgr.es/m/CAOBaU_ZHESFcWva8jLjtZdCLspMj7vqaB2k++rjHLY897ZxbYw@mail.gmail.com
Backpatch-through: 10
src/backend/replication/logical/worker.c