Re: Sequence gaps after restart - Mailing list pgsql-general

From Magnus Hagander
Subject Re: Sequence gaps after restart
Date
Msg-id CABUevExnEa8S+-fN6zRxzE7Jfgq=L=wePnTjTK1hx-QGZdAY3Q@mail.gmail.com
Whole thread Raw
In response to Sequence gaps after restart  (Christopher Sumner <christopher.sumner@adimab.com>)
List pgsql-general
On Tue, May 11, 2021 at 5:55 PM Christopher Sumner
<christopher.sumner@adimab.com> wrote:
>
> Hello.
>
> I'm using PostgreSQL v. 10.14 via AWS Aurora Serverless.  Our users recently noticed gaps in the primary ids for many
ofour tables.  As I understand it, this is due to PostgreSQL effectively caching 32 additional sequence values
(hardcodedin sequence.c ) and a 'restart' occurs.  Losing a few sequence values is normal and generally acceptable.
However,this problem seems to be exacerbated by the scaling events of Aurora Serverless as they require restarts of the
Postgresqlservice. I do not allow AWS AS to force the scaling in our environment. When scaling events happen 10's of
timesper day, the loss is significant. 
>
> Are there any plans to fix this issue or otherwise play nicely with Aurora Serverless?

I don't believe there are any plans to change this in PostgreSQL, as
it's generally not a problem.

AWS Aurora however, is not PostgreSQL, it's a different database
(which shares some parts, but it's fundamentally quite different) It's
also not open source so there is no way for us to know. You'll have to
ask the AWS support about that one.

--
 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/



pgsql-general by date:

Previous
From: Christopher Sumner
Date:
Subject: Sequence gaps after restart
Next
From: Rob Sargent
Date:
Subject: Re: Sequence gaps after restart