Bruce Momjian wrote:
> I know Peter is looking at this, but I looked at and I can't see the
> problem. Every call of exec_prog() that uses pg_resetxlog has
> throw_error = true, and the test there is:
>
> result = system(cmd);
>
> if (result != 0)
> ...
> pg_log(FATAL, ...)
>
> and in pg_log_v() I see:
>
> switch (type)
> ...
> case PG_FATAL:
> printf("\n%s", _(message));
> printf("Failure, exiting\n");
> --> exit(1);
> break;
This was fixed by Peter two days ago in commit
http://git.postgresql.org/pg/commitdiff/ca607b155e86ce529fc9ac322a232f264cda9ab6
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services