Quoth [EMAIL PROTECTED] (Alvaro Herrera):
> On Mon, Dec 06, 2004 at 08:48:04AM -, Julian Scarfe wrote:
>
>> b) Only a dump-restore major version upgrade (which we'll do next time we
>> can take the system out for long enough) will avoid the issue.
>
> "Long enough" could be a minutes or seconds
On Mon, Dec 06, 2004 at 08:48:04AM -, Julian Scarfe wrote:
b) Only a dump-restore major version upgrade (which we'll do next
time we
can take the system out for long enough) will avoid the issue.
On 6 Dec 2004, at 16:18, Alvaro Herrera wrote:
"Long enough" could be a minutes or seconds issue i
On Mon, Dec 06, 2004 at 08:48:04AM -, Julian Scarfe wrote:
> b) Only a dump-restore major version upgrade (which we'll do next time we
> can take the system out for long enough) will avoid the issue.
"Long enough" could be a minutes or seconds issue if you use Slony-I,
I've heard ... (Of cou
From: "Christopher Browne" <[EMAIL PROTECTED]>
> The "empty pages not reclaimed" problem is something that did indeed
> get fixed in the post-7.2 days. I _think_ it was 7.4, but it might
> have been 7.3.
> In short, 7.4.x is indeed a good resolution to your issue.
From: "Tom Lane" <[EMAIL PROTE
Clinging to sanity, [EMAIL PROTECTED] ("Julian Scarfe") mumbled into her beard:
> I've got a box running 7.2.1 (yes, I know :-() in which an index for
> a rapidly turning over (and regularly vacuumed) table is growing
> steadily in size. The index in question is on a timestamp field
> that is just
"Julian Scarfe" <[EMAIL PROTECTED]> writes:
> I've got a box running 7.2.1 (yes, I know :-() in which an index for a
> rapidly turning over (and regularly vacuumed) table is growing steadily in
> size. The index in question is on a timestamp field that is just set to
> now() on the entry of the ro
