Обсуждение: Re: [GENERAL] keeping WAL after dropping replication slots

On 04/04/2017 11:52 PM, Tom DalPozzo wrote:
> Hi,
>
> 2017-04-05 1:55 GMT+02:00 Adrian Klaver <adrian.klaver@aklaver.com
> <mailto:adrian.klaver@aklaver.com>>:
>
>     On 04/04/2017 07:45 AM, Tom DalPozzo wrote:
>
>     Postgres version?
>
> 9.6.1
>
>
>         Hi,
>         I had two replication slots on my primary. Slaves off and
>         (around 800)
>         WALs kept as expected.
>
>
>     Slaves off means?:
>
>
>     You replication set up from the master to the slaves(how many?).
>     Then you disconnected the slaves how?
>
> I have 2 slaves configured with async replication but they were down
>  when I dropped the slots.
>
>     So the 800 WALs number mean you have wal_keep_segments set to 800?
>
> No,  wal_keep_segments is commented.
> 800 is the rough number of files I saw in xlog dir before dropping the
> slots.

What are your settings for?:

archive_mode

archive_command

Do you see anything in the Postgres log that might apply?

>
>
>
>
>         I dropped those slots but over time, the system kept on adding
>         new WALs
>         without reusing them or deleting them.
>         Only after shutdown and restart the system deleted those WAL files.
>         Is that ok?
>         regards
>         Pupillo
>
>
>
>
>     --
>     Adrian Klaver
>     adrian.klaver@aklaver.com <mailto:adrian.klaver@aklaver.com>
>
>
> Regards
> Pupillo
>


--
Adrian Klaver
adrian.klaver@aklaver.com

>             Postgres version?
>
>         9.6.1

Have you considered upgrading to 9.6.2?
There were some fixes, including WAL related:

https://www.postgresql.org/docs/9.6/static/release-9-6-2.html

Not exactly regarding what you see, though...

Bye,
Chris.

On 04/06/2017 11:18 PM, Tom DalPozzo wrote:
>
> Hi,
> 2017-04-06 21:51 GMT+02:00 Adrian Klaver <adrian.klaver@aklaver.com
> <mailto:adrian.klaver@aklaver.com>>:
>
>     On 04/04/2017 11:52 PM, Tom DalPozzo wrote:
>
>         Hi,
>
>         2017-04-05 1:55 GMT+02:00 Adrian Klaver
>         <adrian.klaver@aklaver.com <mailto:adrian.klaver@aklaver.com>
>         <mailto:adrian.klaver@aklaver.com
>         <mailto:adrian.klaver@aklaver.com>>>:
>
>             On 04/04/2017 07:45 AM, Tom DalPozzo wrote:
>
>             Postgres version?
>
>         9.6.1
>
>
>                 Hi,
>                 I had two replication slots on my primary. Slaves off and
>                 (around 800)
>                 WALs kept as expected.
>
>
>             Slaves off means?:
>
>
>             You replication set up from the master to the slaves(how many?).
>             Then you disconnected the slaves how?
>
>         I have 2 slaves configured with async replication but they were down
>          when I dropped the slots.
>
>             So the 800 WALs number mean you have wal_keep_segments set
>         to 800?
>
>         No,  wal_keep_segments is commented.
>         800 is the rough number of files I saw in xlog dir before
>         dropping the
>         slots.
>
>
>     What are your settings for?:
>
>     archive_mode
>
> archive_mode is off
>
>
>     archive_command
>
> it's set as I tested it some months ago but now archive_mode is off
>
>
>     Do you see anything in the Postgres log that might apply?
>
> No, nothing

I am not sure what is going on.

Are the number of WAL files still growing?


--
Adrian Klaver
adrian.klaver@aklaver.com