[
https://issues.apache.org/jira/browse/OAK-10542?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Stefan Egli updated OAK-10542:
------------------------------
Description:
When a document grows too large, part of it is split into previous documents.
That can include revisions of the "_deleted" property.
When reading a node at a particular revision, it first checks whether a node
was deleted at that revision or not. That resolution first checks the revisions
of _deleted locally, i.e. in the main document. Two cases currently exist:
# If it finds a value in the local _deleted map, it accepts that as the valid
one.
# If it doesn't find a value in the local _deleted map, it reads the _deleted
maps from any split documents and searches there for a valid value.
If there was a split of _deleted however, the first case is not sufficient - as
that currently ignores split documents in that case, and the most recent
revision (for a particular read revision) can be in a split doc, rather than
the main doc. This happens if the newer split revisions (of _deleted) are from
a different clusterId than older non-split revisions.
was:
When a document grows too large, part of it is split into previous documents.
That can include revisions of the "_deleted" property.
When reading a node at a particular revision, it first checks whether a node
was deleted at that revision or not. That resolution first checks the revisions
of _deleted locally, i.e. in the main document. Two cases currently exist:
# If it finds a value in the local _deleted map, it accepts that as the valid
one.
# If it doesn't find a value in the local _deleted map, it reads the _deleted
maps from any split documents and searches there for a valid value.
If there was a split of _deleted however, the first case is not sufficient - as
that currently ignores split documents in that case, and the most recent
revision (for a particular read revision) can be in a split doc, rather than
the main doc.
> resolution of _deleted value ignores split docs in certain cases
> ----------------------------------------------------------------
>
> Key: OAK-10542
> URL: https://issues.apache.org/jira/browse/OAK-10542
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: documentmk
> Reporter: Stefan Egli
> Assignee: Stefan Egli
> Priority: Major
>
> When a document grows too large, part of it is split into previous documents.
> That can include revisions of the "_deleted" property.
> When reading a node at a particular revision, it first checks whether a node
> was deleted at that revision or not. That resolution first checks the
> revisions of _deleted locally, i.e. in the main document. Two cases currently
> exist:
> # If it finds a value in the local _deleted map, it accepts that as the valid
> one.
> # If it doesn't find a value in the local _deleted map, it reads the _deleted
> maps from any split documents and searches there for a valid value.
> If there was a split of _deleted however, the first case is not sufficient -
> as that currently ignores split documents in that case, and the most recent
> revision (for a particular read revision) can be in a split doc, rather than
> the main doc. This happens if the newer split revisions (of _deleted) are
> from a different clusterId than older non-split revisions.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
