tag 487983 wontfix thanks Hi Marc,
Marc Cousin [2008-06-25 16:48 +0200]: > /etc/init.d/postgresql-8.3 restart doesn't seem to use -m fast, > whereas /etc/init.d/postgresql-8.3 stop does. So sometimes, on a > very busy database, restart fails... This is actually on purpose. In general it can be assumed that the admin has some control over the situation when doing a restart. I. e. if the server has some ongoing connections/transactions, I'd rather not forcibly terminate it by default (you can still use --force with pg_ctlcluster manually, if you want to). However, on system shutdown we do not have this freedom. We need to guarantee shutdown of the server, but do it as gracefully as possible. Under those conditions, -m fast (and falling back to -m immediate) is appropriate, since it still handles shutdown better than the terminal /etc/init.d/sendsigs, which just kills everything. So unless you have a convincing argument why restart should default to using force as well, I'd rather leave it like it is. What do you think? Thanks, Martin -- Martin Pitt | http://www.piware.de Ubuntu Developer (www.ubuntu.com) | Debian Developer (www.debian.org)
signature.asc
Description: Digital signature
