David Steele

Well, yes, we certainly do that for *backups*, which do not write through temp files. But this is an info file that needs to be *overwritten* and currently this is...

`seconds since epoch` was what I meant. > Do you need to rewrite the info file so many times? Backup and expire are independent processes. Sure it would be possible...

Also important is that this file drives output for the `info` command, so we expect it to be updated when some work is completed. Expire may take a long time,...

As far as I can see, backup.info is saved exactly once per successful backup (https://github.com/pgbackrest/pgbackrest/blob/release/2.38/src/command/backup/backup.c#L2167) and once per expire (https://github.com/pgbackrest/pgbackrest/blob/release/2.38/src/command/expire/expire.c#L1055). If you think there is something different going on you'll...

Ugh, OK, for some reason I thought we were talking about *backup.info*. I'm doing way too many things in parallel today. Yes, the `backup.manifest` is updated periodically during the backup....

We also have a project card to disable this saving if `resume=n` (https://github.com/pgbackrest/pgbackrest/projects/2#card-53604802) but for now `manifest-save-threshold` should work.

Disabling incremental save of manifest.copy when resume=n has been implemented in PR #1759. Final saves are still done via a temp file to maintain atomicity.

> It doesn't have to skip the backup silently, I'd be fine by some skip_backup_on_standby = true option. Yeah, it would need to be a new option. It would not...

Everything starts as a question and we update when we decide for sure that it will be an enhancement. Just never got to this one.

No update. The feature looks viable but we have not had time to schedule it yet. You are welcome to put together a patch yourself.