On Tuesday 01 November 2011 12:00:33 Craig Ringer wrote:
A workaround for reindexing while live is to begin a transaction, create
the new index with a new name, drop the old one, rename the new one to
the old one, and commit. This only requires an exclusive lock for the
period of the drop and
Vincent de Phily vincent.deph...@mobile-devices.fr writes:
The technique kinda works (with some changes) using unique indexes however.
Is
there a functional difference between a unique index and a primary key index
(knowing that my column is not null) ? Or is it just for documentation and
v8.3.4 on linux.
Does reindex table foo require no other users accessing the foo table?
Trying to understand why this seems to be stalled when I attempt this on a live
DB (if runs fine/fast on a copy of the DB that no one uses).
Should I run this inside a transaction?
Thanks in Advance !
On 01/11/11 02:51, Gauthier, Dave wrote:
v8.3.4 on linux.
Does reindex table foo require no other users accessing the foo
table? Trying to understand why this seems to be stalled when I attempt
this on a live DB (if runs fine/fast on a copy of the DB that no one uses).
Yes, it
Craig Ringer ring...@ringerc.id.au writes:
On 01/11/11 02:51, Gauthier, Dave wrote:
Does reindex table foo require no other users accessing the foo
table? Trying to understand why this seems to be stalled when I attempt
this on a live DB (if runs fine/fast on a copy of the DB that no one