Ticket #26 (closed enhancement: wontfix)

Opened 3 years ago

Last modified 3 years ago

Metadata from metadata servers should only be downloaded when needed

Reported by: jcnelson Owned by: jcnelson
Priority: critical Milestone:
Component: component1 Version:
Keywords: Cc:

Description (last modified by jcnelson) (diff)

Instead of having a metadata server download and then validate the metadata from another metadata server, and keep all of the remote server's metadata locally, just verify that the remote metadata server is still alive.

POSSIBLE IMPLEMENTATION:

  • A remote metadata server is represented in the master copy by a specially-named file .$metadata$ in an (empty) directory. That file is a symlink that links to the URL of the metadata server.
  • When mdcgi generates metadadata, it looks for the .$metadata$ file in the directory. If it exists, then it pairs that directory with the URL that .$metadata$ links to.
  • The client interprets a directory paired with a URL as a metadata server--the client resolves the metadata server, gets its metadata, and populates that directory with it.

Change History

comment:1 Changed 3 years ago by jcnelson

  • Description modified (diff)

comment:2 Changed 3 years ago by jcnelson

  • Status changed from new to closed
  • Resolution set to wontfix

Metadata servers are unlikely to ever crawl other metadata servers anyway--this is not a use-case we should promote. If a client wants a filesystem composed of multiple metadata sources, (s)he should mount multiple Syndicate clients. Close as wontfix.

Note: See TracTickets for help on using tickets.