From b17996c4931f7e74b74b8df05dacbbffb69064cf Mon Sep 17 00:00:00 2001 From: David Turner Date: Fri, 18 Oct 2024 04:37:07 +0100 Subject: [PATCH] Reword docs on snapshot repo backup Because of #93575 it's not sufficient to mark repositories with `readonly: true` while taking a backup. The only safe way to avoid writes is to completely unregister them. --- .../snapshot-restore/register-repository.asciidoc | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/docs/reference/snapshot-restore/register-repository.asciidoc b/docs/reference/snapshot-restore/register-repository.asciidoc index 2147ad3c684f3..6c1319c2c71b1 100644 --- a/docs/reference/snapshot-restore/register-repository.asciidoc +++ b/docs/reference/snapshot-restore/register-repository.asciidoc @@ -248,10 +248,11 @@ that you have an archive copy of its contents that you can use to recreate the repository in its current state at a later date. You must ensure that {es} does not write to the repository while you are taking -the backup of its contents. You can do this by unregistering it, or registering -it with `readonly: true`, on all your clusters. If {es} writes any data to the -repository during the backup then the contents of the backup may not be -consistent and it may not be possible to recover any data from it in future. +the backup of its contents. If {es} writes any data to the repository during +the backup then the contents of the backup may not be consistent and it may not +be possible to recover any data from it in future. Prevent writes to the +repository by unregistering the repository from the cluster which has write +access to it. Alternatively, if your repository supports it, you may take an atomic snapshot of the underlying filesystem and then take a backup of this filesystem