You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is correct for the govuk which uses external versioning, but incorrect for mainstream, government and detailed, which use internal versioning (the default).
This doesn't cause any race conditions in those indices because we don't use optimistic locking when updating the non-govuk indices anyway.
But it is confusing, because it implies that we use versioning in non-govuk indices when we don't. This has slowed down our investigation of search duplicates because we weren't sure if the mix of versioning systems could cause duplicates of documents. (It can't, as far as we know, but we wasted time investigating it.)
From: https://trello.com/c/xiYv8sH9/38-fix-versiontype-in-popularity-update
We set
'_version_type' => 'external_gte'
on every document when updating the popularity data: https://github.com/alphagov/rummager/blob/master/lib/govuk_index/popularity_worker.rb#L23This is correct for the govuk which uses external versioning, but incorrect for mainstream, government and detailed, which use internal versioning (the default).
This doesn't cause any race conditions in those indices because we don't use optimistic locking when updating the non-govuk indices anyway.
But it is confusing, because it implies that we use versioning in non-govuk indices when we don't. This has slowed down our investigation of search duplicates because we weren't sure if the mix of versioning systems could cause duplicates of documents. (It can't, as far as we know, but we wasted time investigating it.)
More on ES versioning: https://www.elastic.co/blog/elasticsearch-versioning-support
The text was updated successfully, but these errors were encountered: