Besides metadata records, an OpenGeoMetadata repository can contain a few files at its root that describe the repository itself. Harvesters skip these files when indexing records.
File
Purpose
withdrawn.json
Lists records the repository has withdrawn, so harvesters can delete them. See below.
layers.json
Legacy. Maps record identifiers to their folders, for people browsing the repository. See Share on OpenGeoMetadata.
A log of records that the repository has deliberately stopped publishing. Harvesters read it and delete the listed records from their indexes. For when and how to use it, see Withdraw Records.
{"$schema":"https://opengeometadata.org/schema/ogm-withdrawals-1.0.json","ogm_version":"1.0","withdrawn":[{"id":"stanford-bb058zh0946","date":"2026-03-14","reason":"superseded","note":"Replaced by the 2026 parcel release.","is_replaced_by":["stanford-cc123dd4567"]},{"id":"stanford-qp039xv4451","date":"2026-03-14","reason":"upstream-removed"}]}
An array with one entry per withdrawn record. Each entry has the fields below.
The rules harvesters follow:
A record that is still published is never deleted. If an id is listed in withdrawn.json and also appears in a record in any harvested repository, the record stays and the harvester warns. A mistaken entry is therefore harmless.
Each id counts once. If an id is listed more than once, the last entry's date, reason, note, and replacements apply. The order of entries never changes whether records are withdrawn.
Unknown properties are ignored, both on the file and on entries.
Entries are permanent. Harvesters read the whole log on every run, and one that hasn't run in a while learns about a withdrawal only if the entry is still there. Remove an entry only when you publish the record again.
To identify the withdrawn record so that harvesters can delete it from their indexes.
Entry Guidelines
Enter the record's identifier exactly as it appeared in the record: the value of ID (id) for OGM Aardvark records, or layer_slug_s for GBL 1.0 records. Do not add leading or trailing spaces.
Commentary
This is the value a harvester deletes from Solr, so it must match character for character. It is not a druid, ARK, file path, or the value of Identifier. If your repository published the same item under different identifiers in more than one schema (for example, in both metadata-1.0/ and metadata-aardvark/), add an entry for each.
Enter a date (2026-03-14), or a date and time with a UTC offset or Z (2026-03-14T09:30Z, 2026-03-14T09:30:00-07:00). Seconds and fractional seconds are optional.
Commentary
Follows the W3C Date and Time Format. A time without a UTC offset, a year on its own, and a year and month on their own are not valid. The date is for audit and reporting; harvesters do not use it to decide what to delete.
To explain the withdrawal to whoever reads the log or a harvest report.
Entry Guidelines
Enter a short free-text explanation.
Commentary
Notes are not meant for display to end users. Because withdrawn.json is public and kept in git history, do not include personal information, legal detail, or the identity of anyone who requested the withdrawal.
To point to the record(s) that replace the withdrawn one.
Entry Guidelines
Enter the ID of each replacement record.
Commentary
Mirrors Is Replaced By (dct_isReplacedBy_sm). If you are keeping the old record published, don't withdraw it. Instead, list it in Replaces on the newer record, which is the field GeoBlacklight uses to link the two.