Enable PDF Conversion Debug Logging
Steps
The Docker converter runs pdflatex inside a container and keeps to itself everything that run printed. A PDF conversion that fails reports those compilation logs with its error, but one that succeeds and hands you a PDF you did not expect reports nothing at all. Raising a single logger to DEBUG puts the whole compilation output in the job log of every export.
- Open Administration > Logging, the section the Logging application adds. Filter the Logger column on org.xwiki.contrib.latex.internal.pdf, choose DEBUG in the Actions column of that row, and click Set.
Set the level on that logger rather than on one of the rows below it. Every logger nested under it inherits the level, so this one covers the converter, the container it drives, and anything else the conversion goes through.
- Check that the Level column of that row now reads DEBUG.
The new level applies at once, with nothing to restart. It is a runtime setting, though: XWiki comes back at the default level, so you have to set it again after each restart.
- Export a page to PDF (LaTeX) and open the Job log of the finished export. The pdflatex run is in it line by line, exactly as the container printed it.
FAQ
The output is not in the server log. Where does it go?
An export runs as a job, and XWiki collects what a job logs into that job's own log instead of sending it to the server console. The converter's output is therefore on the export page, under Job log, and nowhere else.
I use the local process converter and DEBUG changes nothing. Why?
Because there is nothing to add. The local process converter logs its pdflatex output at the INFO level, so every export already carries it in its job log, and a command that fails reports its output with the error. This procedure is for the Docker converter.
Can I leave the logger at DEBUG?
Yes, until the next restart resets it. Each PDF export then adds a few hundred lines of compilation output to its own job log.
