This issue appears to be related to this bug: Record not processed because submitted version: 1643750256 is less or equal to previously submitted version {1} (403 Forbidden) Posting evidence here rather than following up with crossref to fix the symptom because it looks like some bad data being generated by OJS.
We’re running 3.3.0.14 with all plugins up to date (except ORCID, which was only updated a couple of weeks ago)
We’ve seeing oddities with crossref.
(a) Some DOIs point to our test server. For example: https://ojs.victoria.ac.nz/ajl/article/view/1765 has DOI 10.26686/ajl.v2i0.1765 / https://doi.org/10.26686/ajl.v2i0.1765 which points to our test server
https://ojs-test.vuw.ac.nz/ajl/article/view/1765 . This MAY be an issue at our end and be completely unrelated (but if so, speaks to the need to better check that DOIs aren’t pushed from test servers).
(b) Some of the timestamps in crossref appear to be in the wrong format, for example:
That certainly looks like a mangled timestamp there, the crossref schema says YYYYMMDDHHMMssvvv, but thats, a raw unix timestamp (seconds since 1 Jan 1970). Probably needs better sanity checking on the crossref side and a fix for the crossref plugin.
(c) We’re seeing the error described in Record not processed because submitted version: 1643750256 is less or equal to previously submitted version {1} (403 Forbidden) when we try to push publish new DOIs.
cheers
stuart