Postgres Pro
Downloads
Home   >   mailing lists

Thread: PostgreSQL Cascade streaming replication problem

PostgreSQL Cascade streaming replication problem

From
Jakub Janeček
Date:
Hi everyone,

i have one master and two slaves - cascading replication. Master -> Slave1 -> Slave2

Slave2 crashed two days ago. Problem is on Slave1, where are comulating WAL files. Archive mode is turn off. There is about 6.000 WAL Files, but in folder archive_status are files with .done name.

How can I remove WAL files, which are "waiting" for Slave2? I changed configration on Slave1 (i commented this options "wal level = hot_standby" and "max_wal_senders = 1"). But without effect. 

What did i do wrong? I need stop comulating WAL files and remove old WAL files, which are processed and are only "waiting"..

Thank you for your tips.
Regards,

Jakub

Re: PostgreSQL Cascade streaming replication problem

From
Michael Paquier
Date:
On Sun, Apr 01, 2018 at 06:26:51PM +0000, Jakub Janeček wrote:
> What did i do wrong? I need stop comulating WAL files and remove old WAL
> files, which are processed and are only "waiting"..

Perhaps wal_keep_segments is set and you forgot about it or you used a
replication slot that you forgot to drop on slave 1?
--
Michael
Attachment

Re: PostgreSQL Cascade streaming replication problem

From
Jakub Janeček
Date:
I have set up wal_keep_segments = 100 on Slave1, but no effect. I tried comment wal_keep_segments, no effect too... Actually i have 8.500 WAL Files, disk space is low. I tried drop replication slot on Slave1, no effect. Is possible to manually remove WAL files? 

Thanks

Jakub

2018-04-03 4:43 GMT+02:00 Michael Paquier <michael@paquier.xyz>:
On Sun, Apr 01, 2018 at 06:26:51PM +0000, Jakub Janeček wrote:
> What did i do wrong? I need stop comulating WAL files and remove old WAL
> files, which are processed and are only "waiting"..

Perhaps wal_keep_segments is set and you forgot about it or you used a
replication slot that you forgot to drop on slave 1?
--
Michael

Re: PostgreSQL Cascade streaming replication problem

From
Jakub Janeček
Date:
Hi everyone,

problem is solved. 

There was a delay between setup (drop replication slot) and execution.

Jakub

2018-04-03 4:43 GMT+02:00 Michael Paquier <michael@paquier.xyz>:
On Sun, Apr 01, 2018 at 06:26:51PM +0000, Jakub Janeček wrote:
> What did i do wrong? I need stop comulating WAL files and remove old WAL
> files, which are processed and are only "waiting"..

Perhaps wal_keep_segments is set and you forgot about it or you used a
replication slot that you forgot to drop on slave 1?
--
Michael