The CMIS change log currently uses a change log token that is the log event time, in milliseconds.
In the very rare case when all the events returned by a page of getContentChanges all have the exact same event time, the next change log token will still be identical to the previous one so the next page returned will be at the same point in the audit thus causing an infinite loop for a client that requests all content changes.
To avoid this we must switch to using the native event id and not an event date. Drive already does this correctly.
- depends on
-
NXP-20269 CMIS change log doesn't work with Elasticsearch audit
- Resolved