Justin Pasher <justinp@xxxxxxxxxxxxxxxxxxx> writes: > A client of ours has requested that we change the time on the server to > match their time zone (one hour ahead of us). Are there any know issues > or gotchas to look out for when bumping the server time forward an hour? > I couldn't imagine any issues occurring, since it's really just like the > server when 1 hour without performing any activity. Obviously any fields > in the database stored with a time without a timezone would be off > afterwards, but that's not a huge concern. > Another client has requested the same thing, but their time is two hours > behind us. This is the change that I'm more worried about, since you are > going to have some data/time overlap on the server after the change is > made. Are there any known issues with this situation in Postgres? Should > everything be shutdown prior to the change, then restarted again? There shouldn't be any need to change the server's clock at all, at least not if you are running a sane operating system that keeps the underlying time in GMT. What you should be doing is setting the default value of PG's TimeZone parameter to match what these clients want. Or suggest to them that they set it themselves on a per-session or per-user basis. regards, tom lane
