Hi,
I'm talking about our own massively bloated toast table - described in
an earlier post - that I think I can replicate. I didn't mean to
steal your thread, but the problem seems very similar, and we're using
9.1. I don't know a lot about Postgres internals, but to me it smells
like a bug of some sort.
On Mon, Jul 15, 2013 at 7:23 PM, Bradley McCune <bradley.mccune@noaa.gov> wrote:
> David,
>
> I'm sorry, but I'm not sure that I follow how this is pertinent to this
> particular thread. Are you proposing a way to replicate the scenario we
> experienced of our massively bloated TOAST table? If so, I'm not entirely
> sure that's doable given that the source of the issue was never clear.
> There still remains a number of reasons for why that table had so much
> "still in use" bloat. At this moment, it's near impossible to tell given
> that it is no longer a problem.
>
> Thanks for the offer, and I apologize if I'm just slightly ignorant about
> your intentions.
>
>
> On Mon, Jul 15, 2013 at 4:33 AM, David Welton <davidw@dedasys.com> wrote:
>>
>> Hi,
>>
>> I think I could write a script to do something similar to what is
>> happening if anyone is interested. I'd want some direction as to the
>> best way to handle this though: it'd be easier for me to script it as
>> Rails code because that's what the app is. Perhaps from that we can
>> get the generated SQL so as to make it easier for others to deal with.
>> The operation itself is basically:
>>
>> * Extract a value from a row of a table that is stored as a bytea.
>>
>> * Unmarshall it into a Ruby object.
>>
>> * Add to that Ruby object.
>>
>> * update the row and set the value by marshalling the Ruby object.
>>
>> I suspect that the actual value isn't terribly relevant, and they
>> how's and why's of what it is like it is are best left for a different
>> discussion.
>>
>> --
>> David N. Welton
>>
>> http://www.dedasys.com/
>
>
>
>
> --
> Bradley D. J. McCune
--
David N. Welton
http://www.dedasys.com/