Skip to content

Detecting exising indices and shards is broken #13265

@s1monw

Description

@s1monw

Today we rely on FS operations to find the indices on disk or to find the shards for an index. This is super error prone and requires String parsing. We should on each level index -> shard know exactly what to expect and don't use directory listings which are expensive and subject to change. There should be a metadata file on each level that is atomically written that we open and see what we have to expect no matter of what's on the FS.

Metadata

Metadata

Assignees

No one assigned

    Labels

    :Distributed/EngineAnything around managing Lucene and the Translog in an open shard.>bugdiscuss

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions