On 11/7/22 16:03, da avory wrote:
Hi all,
not sure whether this is the right list to ask. I do have two (small!?)
feature suggestions, both regarding backup and replication, which might
appeal to a wider audience.
1) it would be very great to have the option to add a timestamp to all
lines of output of pg_dump and pg_restore. We regularly do dump/restores
that take multiple days
Are you using "--jobs=X --format=directory", or old school plain text dumps?
and this would help immensely to see bottlenecks, time distribution and
get an overall feel for which steps take how long.
While timestamps would definitely be useful, the --verbose option with
stdout *and* stderr redirected to a file plus some manual effort will answer
your questions.
2) It would be equally great to be able to retrieve the actual value of
recovery_min_apply_delay config parameter that is configured on a
streaming client as a value on the primary (as a field in
pg_stat_replication). Ideally in a consistent form, i.e. converted to
seconds or interval. We currently have our ops team monitoring our
streaming clients in various monitoring views and while we can of course
see pending transfer and replay volume there is no way to see the intended
apply_delay without storing it somewhere on the primary (where it will get
inevitably out of sync with the used value).
Many thanks
Dean
--
Angular momentum makes the world go 'round.
