On Thu, Jul 28, 2016 at 3:16 PM, Bruce Momjian <bruce@momjian.us> wrote:
> On Thu, Jul 28, 2016 at 12:35:23AM -0700, Jeff Janes wrote:
>> On Wed, Jul 27, 2016 at 9:48 PM, John R Pierce <pierce@hogranch.com> wrote:
>> > On 7/27/2016 9:39 PM, Jeff Janes wrote:
>> >>
>> >> That depends on how how many objects there are consuming that 1 TB.
>> >> With millions of small objects, you will have problems. Not as many
>> >> in 9.5 as there were in 9.1, but still it does not scale linearly in
>> >> the number of objects. If you only have thousands of objects, then as
>> >> far as I know -k works like a charm.
>> >
>> >
>> > millions of tables?
>>
>> Well, it was a problem at much smaller values, until we fixed many of
>> them. But the perversity is, if you are stuck on a version before the
>> fixes, the problems prevent you from getting to a version on which it
>> is not a problem any more.
>
> Uh, that is only true if the slowness was in _dumping_ many objects.
> Most of the fixes have been for _restoring_ many objects, and that is
> done in the new cluster, so they should be OK.
There have been improvements on both sides. For the improvements that
need to exist in the old-server to be effective, we did backpatch the
main one back to 9.1, in the October 2015 releases, specifically to
help people get off the old versions. So if you are on 9.1 with
tens/hundreds of thousands of objects, you need to do a minor version
upgrade to at least 9.1.19 before doing the major version upgrade. If
you are on 9.0 or before with so many objects, you don't have a lot of
good options.
Cheers,
Jeff
