Greetings,
* Andrew Dunstan (andrew.duns...@2ndquadrant.com) wrote:
> On 2/22/19 2:15 PM, Andres Freund wrote:
> > On 2019-02-21 16:27:06 +0100, Andreas Karlsson wrote:
> >> I have not heard many requests for bringing back the old behavior, but I
> >> could easily have missed them. Either way I do
On 2/22/19 2:15 PM, Andres Freund wrote:
> Hi,
>
> On 2019-02-21 16:27:06 +0100, Andreas Karlsson wrote:
>> I have not heard many requests for bringing back the old behavior, but I
>> could easily have missed them. Either way I do not think there would be much
>> demand for an in-place VACUUM FUL
Hi,
On 2019-02-21 16:27:06 +0100, Andreas Karlsson wrote:
> I have not heard many requests for bringing back the old behavior, but I
> could easily have missed them. Either way I do not think there would be much
> demand for an in-place VACUUM FULL unless the index bloat problem is also
> solved.
Thanks for getting back to me. I had a small discussion with @sfrost on the
slack team and understand the issue better now. I must admit I didn't
realize that the scope of WAL extended to VACUUM operations which is why I
suggested a new journaling system. I realize now the issue is not safety(as
th
On 2/21/19 12:16 AM, Ryan David Sheasby wrote:
I was reading on VACUUM and VACUUM FULL and saw that the current
implementation of VACUUM FULL writes to an entirely new file and then
switches to it, as opposed to rewriting the current file in-place. I
assume the reason for this is safety in the
