Fix issues with FAILOVER
authorSteve Singer <ssinger@ca.afilias.info>
Mon, 10 Apr 2017 13:38:15 +0000 (09:38 -0400)
committerSteve Singer <ssinger@ca.afilias.info>
Mon, 10 Apr 2017 13:38:15 +0000 (09:38 -0400)
commit0974d0ca93eff1ed819beffa13370f1aff81e047
treedcd1a1fa8b0efc7edbdc117081569672a14cd137
parent9a4a56c39ba2a95592c74a49527bd8fca1da9e7d
Fix issues with FAILOVER
In particular it is possible when doing a multi-node failover
with a cascaded node for the result after the failover to
have nodes with a self subscription, a row in sl_subscribe
with the provider and receiver equal.

In fixing that issue we also discovered that slon was not
reloading the set origin's into memory following a FAILOVER
command.  This could mean that we were processing SYNC events
from a node which without realizing that node now is the origin
for a set.
src/backend/slony1_funcs.sql
src/slon/remote_worker.c
src/slon/runtime_config.c
src/slon/slon.h