On Mon, Feb 10, 2020 at 9:41 AM Fujii Masao <masao.fu...@oss.nttdata.com> wrote: > > > > On 2020/02/10 17:23, Sergei Kornilov wrote: > > Hello > > > > Seems bug was introduced in caba97a9d9f4d4fa2531985fd12d3cd823da06f3 - in > > HEAD only > > > > In REL_12_STABLE we have: > > > > bool is_recovery_guc_supported = true; > > > > if (PQserverVersion(conn) < MINIMUM_VERSION_FOR_RECOVERY_GUC) > > is_recovery_guc_supported = false; > > > > snprintf(filename, MAXPGPATH, "%s/%s", basedir, > > is_recovery_guc_supported ? "postgresql.auto.conf" : > > "recovery.conf"); > > > > cf = fopen(filename, is_recovery_guc_supported ? "a" : "w"); > > > > It looks correct: append mode for postgresql.auto.conf > > > > In HEAD version is_recovery_guc_supported variable was replaced to inversed > > use_recovery_conf without change fopen mode. > > Yes! Thanks for pointing out that! > So the patch needs to be applied only in master.
+1. We should absolutely not be overwriting the auto conf. -- Magnus Hagander Me: https://www.hagander.net/ Work: https://www.redpill-linpro.com/
