Indexing is pausing on long index run

[henning-gerhardt]

As indexing of million entries takes a few days you are closing the browser or browser tab as you must or you need to do.

If you are closing the browser or browser tab then indexing is still running for next 24 or 36 hours (guessing!) but then indexing is pausing. If you are login back after 2 or 3 days then indexing is resuming its job but not in any case with all remaining entries. Sometimes indexing is skipping some entries or a whole type of objects (f.e. all projects).

Indexing is not paused if you have the possibility to let the browser tab open and refreshing the browser tab after a few hours of inactivity (12 hours or so, just guessing).

[henning-gerhardt]

An additional note: there are not log messages in Kitodo log (log level info) nor catalina.out files why parts of indexing are skipped on resuming after login.

[matthias-ronge]

Maybe related to #3719

[matthias-ronge]

Maybe related to #4947

[matthias-ronge]

This happens because the currentIndexState in the IndexingService is not reset in runIndexing() after the end of the indexing thread (after indexerThread.join()), but in the function getProgress(), so only when the frontend polls the progress. If no browser window performs the query, indexing will wait indefinitely for the previous indexing step to be released.