Thread: WAL-dir filling up with wal_level = logical
Version: PG-10.4
I don't (yet) have any replication-slots configured, and hence no standbys are subscribed, but I have wal_level = logical configured to be able to add subscribers later. I'm seeing that WAL-dir is filling up with WAL-files (now 17GB and not declining), is this expected behaviour?
Thanks.
--
Andreas Joseph Krogh
Andreas Joseph Krogh
På fredag 03. august 2018 kl. 12:12:33, skrev Andreas Joseph Krogh <andreas@visena.com>:
Version: PG-10.4I don't (yet) have any replication-slots configured, and hence no standbys are subscribed, but I have wal_level = logical configured to be able to add subscribers later. I'm seeing that WAL-dir is filling up with WAL-files (now 17GB and not declining), is this expected behaviour?Thanks.
Hm, a co-worker noticed that we have max_wal_size = 16GB, that might be the reason:-)
--
Andreas Joseph Krogh
Andreas Joseph Krogh
On 3 August 2018 12:12:33 CEST, Andreas Joseph Krogh <andreas@visena.com> wrote: >Version: PG-10.4 > >I don't (yet) have any replication-slots configured, and hence no >standbys are >subscribed, but I have wal_level = logical configured to be able to add > >subscribers later. I'm seeing that WAL-dir is filling up with WAL-files >(now >17GB and not declining), is this expected behaviour? > No. Maybe a not-working archive_command? Regards, Andreas -- 2ndQuadrant - The PostgreSQL Support Company
On 3 August 2018 12:33:26 CEST, Andreas Joseph Krogh <andreas@visena.com> wrote: >På fredag 03. august 2018 kl. 12:12:33, skrev Andreas Joseph Krogh < >andreas@visena.com <mailto:andreas@visena.com>>: >Version: PG-10.4 > >I don't (yet) have any replication-slots configured, and hence no >standbys are >subscribed, but I have wal_level = logical configured to be able to add > >subscribers later. I'm seeing that WAL-dir is filling up with WAL-files >(now >17GB and not declining), is this expected behaviour? > >Thanks. > >Hm, a co-worker noticed that we have max_wal_size = 16GB, that might be >the >reason:-) > Yes ;-) Regards, Andreas -- 2ndQuadrant - The PostgreSQL Support Company
On Fri, Aug 03, 2018 at 12:40:16PM +0200, Andreas Kretschmer wrote: > On 3 August 2018 12:33:26 CEST, Andreas Joseph Krogh <andreas@visena.com> wrote: >>Hm, a co-worker noticed that we have max_wal_size = 16GB, that might be >>the >>reason:-) > > Yes ;-) Worth mentioning that this is a soft size, and not a hard size, hence depending on your worload you may see more WAL segments than what is set in 16GB. The docs mention that, so no need to be surprised. -- Michael