Re: replication lag despite corrective config

On 11/19/18 11:09 PM, Laurenz Albe wrote:
> With these settings, any conflicting query will be canceled after five
> minutes. Perhaps your actual settings are different.
>
> What do you get for
>
> SELECT * FROM pg_settings WHERE name = 'max_standby_streaming_delay';
Hi Laurenz, thanks for backing up my understanding.

The settings are as I reported:

db=> show max_standby_streaming_delay;
 max_standby_streaming_delay
-----------------------------
 5min

db=> show hot_standby_feedback;
hot_standby_feedback
----------------------
 on

db=> show max_standby_archive_delay;
max_standby_archive_delay
---------------------------
 30s

db=> select * from pg_settings where name = 'max_standby_streaming_delay';
-[ RECORD 1 ]---+----------------------------------------------------------
name            | max_standby_streaming_delay
setting         | 300000
unit            | ms
category        | Replication / Standby Servers
short_desc      | Sets the maximum delay before canceling queries when a
<snip>
extra_desc      | <<NULL>>
context         | sighup
vartype         | integer
source          | configuration file
min_val         | -1
max_val         | 2147483647
enumvals        | <<NULL>>
boot_val        | 30000
reset_val       | 300000
sourcefile      | <<NULL>>
sourceline      | <<NULL>>
pending_restart | f

In case it's of additional interest, the full configs for the replica
and the primary are here:

https://gist.github.com/wkalt/a33e7628c63bfc71a6d0ee85103302d9

and here: https://gist.github.com/wkalt/a9dd5ceb0b01b50615c51063ee972117.