Troubleshooting
Thesis harvest looks failed
What to do when a thesis harvest run ended with the Failed status or its results list problems, by waiting, asking the repository's administrator or starting the harvest again, depending on the cause.
When a run cannot get the list of records from the repository, or stops halfway, it ends with the Failed status. A problem in a single thesis does not stop the run; it only appears in the list of problems. Nor does a page the repository cannot produce. To see the problems, press the run's Details button in the Recent harvests table. Follow the section below that matches the cause the problem line gives.
The repository is not answering
If the line says "The source archive could not be reached." or that the repository is not answering right now, the repository did not answer despite several attempts. Start the run again a little later. If you do not start it by hand, the system tries again the next night. If the problem lasts for several days, contact the repository's administrator.
The repository gave an answer that cannot be read
If the line says "The source archive sent an answer that cannot be read.", or that the repository reported an OAI error or answered unexpectedly, the repository's answer could not be read. Try again a little later. If the problem continues, pass the error code on the line to the repository's administrator.
robots.txt no longer allows it
If the line says "The source archive's robots.txt does not allow its OAI-PMH address.", the repository's robots.txt changed after the check. Ask the repository's administrator to open the OAI-PMH address again.
The run stopped halfway
If the line says "The harvest stopped without finishing.", the run took longer than twelve hours and the system closed it as failed. Start the run again. The theses read until then stay in the catalogue, and the unchanged ones are not written again.
If the line says "The archive's address, format or sets changed during the harvest. Start it again.", the repository's address, the record format or the sets were changed while the run was going on. Start the run again with the new settings.
Problem lines in a finished run
A run with the Done status can also list the problems of single records.
- If the line says the record is not a thesis, the record is of another kind, such as a bachelor's thesis or a dataset, and is not harvested. If it is a thesis, correct its type in the repository.
- "The record has no title." says the record has no title in any language. Add the title in the repository.
- "The record's OAI identifier is longer than 128 characters." says the record's identifier is too long to be harvested. Tell the repository's administrator.
- "The record could not be added to the catalogue." says the catalogue refused the record. The rest of the message gives the reason.
- "The source archive could not produce this page, so the set's records were asked for one at a time." says the repository could not produce a page in spite of several tries. It usually fails because it cannot write one record on that page. JournTech then gets the set's list again from its start and asks for the records it has not harvested yet one at a time. The run therefore takes longer than usual.
- If the line says the repository could not produce the record, the repository refused the record with an error each time it was asked for, and the record was not harvested. The error usually comes from the record itself; its abstract may hold a character that XML does not accept, for example. Pass the record identifier on the line to the repository's administrator. Once the record is corrected in the repository, the next sync brings it.
- If the line says the thesis stays a draft, see A harvested thesis stays a draft.
Full-text problems
These problems only come up while full texts are copied.
- If the line says the full text was not downloaded for now, the repository did not give the file at that moment. The file is asked for again up to five times, twenty minutes apart.
- If the line says the full text was not downloaded, the repository refused to give the file. Check that the file is open access in the repository.
- "The full text is larger than the 150 MB limit." says the file is too large to be copied. In that case the thesis page does not show the full text.
The removal check did not remove theses
If the line says "Many theses were missing from the source archive's list. As that is beyond the safety limit, no thesis was unpublished.", see The removal check did not remove the theses.
The alert to the system administrators
Every run that ends as failed sends an operations alert to the system administrators. At most one alert is sent per hour. The alert links to the Thesis harvest screen. Alerts are explained in Operations Alerts.