Bugs in superuser_reserved_connections and max_wal_senders vs max_connections
От | Magnus Hagander |
---|---|
Тема | Bugs in superuser_reserved_connections and max_wal_senders vs max_connections |
Дата | |
Msg-id | CABUevEwm4KjR6N1O4N9_SU0Tkxujigj7oO2Ug2OvbXo_G9EFNQ@mail.gmail.com обсуждение исходный текст |
Ответы |
Re: Bugs in superuser_reserved_connections and max_wal_senders vs max_connections
|
Список | pgsql-hackers |
Both superuser_reserved_connections and max_wal_senders can be set to a value that's higher than max_connections, which is quite pointless, and annoying :) The docs for superuser_reserved_connections say "The value must be less than the value of max_connections.", but this is never enforced. Well, it's strangely enfocred. With max_connections=100, I can set superuser_reserved_connections to 103. Looks like this is caused by autovacuum_max_workers - because if I increase that one to 4, then I can get it up to104... The check in PostmasterMain():if (ReservedBackends >= MaxBackends){ write_stderr("%s: superuser_reserved_connections mustbe less than max_connections\n", progname); ExitPostmaster(1);} should probably check against MaxConnections instead of MaxBackends, I think? The docs for max_wal_senders say nothing at all about the relation to max_connections, which it really should. And it should probably be enforced the same way that superuser_reserved_connections is, so you can't set it to silly high values. -- Magnus HaganderMe: http://www.hagander.net/Work: http://www.redpill-linpro.com/
В списке pgsql-hackers по дате отправления: