Nikolay Antonov

|approach|backup|backup time|restore time| | --- | --- | --- | --- | |SplitMerge x16| stream_20240330T172542Z | 22m26.408s | 16m57.928s | | FIFO x16 | stream_20240330T165057Z | 23m30.876s | 14m11.283s |...

Hi, @catalogoanatom **MySQL configuration** In order to achieve quite reliable PiTR please turn on GTIDs on mysql itself. Also I would recommend to use row-based replication (as more deteministic). Don't...

Hi, @Deracination Do you have binlogs uploaded? Technically, it is ok. You should be able to restore MariaDB from backup without any issues. However Point in Time Recovery will not...

Hi @catalogoanatom ! I don't have configured MySQL nearby. But I used to use '%hostname%-bin-00001' file names. This will prevent binlog overrides. It seems that it is enough to set...

## Current behavior reasoning You are right - current binlog never backed up and this can lead to lost transactions. However, in my practice, it is rarely the case for...

@terranus which version of WAL-G do you use?

At first glance it looks like an a bug - we close file twice: [here](https://github.com/wal-g/wal-g/blob/master/internal/splitmerge/mergewriter.go#L65) and [here](https://github.com/wal-g/wal-g/blob/master/internal/stream_fetch_helper.go#L73) Need some time to review this code one more time. I don't think...

> After that, nothing changed, there are still 18 backups (maybe because of the "dry run"? but why?). By default `wal-g delete` doing dry-run. In order to perform deletion -...

We cannot use binlog from `xtrabackup_info` because binlogs are uploaded from master host, and backups can be made from any host in the cluster, so there is chance that binlog...

I just realized that the suggested approach can work only in scenarios when you making database backups from the primary. In this case binlog_name + position will match values that...