ecf97e9415
* We now have concurrent repository operations so the one at a time limit does not apply any longer * Initialization was never slow solely due to loading information about all existing snaphots (though this contributed) but also because two cluster state updates and a few writes to the repository had to happen before initialization could return * Repo data necessary for a snapshot create operation is now cached on heap so loading it is effectively instant * Snapshot initialization is just a single CS update now * Initialization does no writes to the repository whatsoever * Fixed missing `repository` |
||
|---|---|---|
| .. | ||
| apis | ||
| index.asciidoc | ||
| monitor-snapshot-restore.asciidoc | ||
| register-repository.asciidoc | ||
| restore-snapshot.asciidoc | ||
| take-snapshot.asciidoc | ||