On Mon, Apr 16, 2018 at 12:21 PM, Tom Lane <t...@sss.pgh.pa.us> wrote:
> Keith Fiske <keith.fi...@crunchydata.com> writes: > > On Mon, Apr 16, 2018 at 12:09 PM, Tom Lane <t...@sss.pgh.pa.us> wrote: > >> This is not a great idea, no. You could be getting strange misbehaviors > >> in e.g. string comparison, because strcoll() will expect UTF8 data and > >> will likely not cope well with data that isn't valid in that encoding. > > > And pg_controldata was able to show that the CTYPE and COLLATE were UTF8 > on > > the old system. If that's the case, do you still think it's a good idea > to > > set the COLLATE and CTYPE to "C"? > > Well, if the customer's been happy with the behavior of the system so far, > maybe it's all right. But this is sure the first thing I'd look at if > there are any gripes about its behavior with non-UTF8 strings. I'd be > especially worried about this if you try to migrate the database to any > new platform, as it's a bet about the behavior of libc not PG itself. > > regards, tom lane > This is going from RHEL 6.7 to RHEL 7.4 It is a dump and restore upgrade as well. -- Keith Fiske Senior Database Engineer Crunchy Data - http://crunchydata.com