Using HTML headers and footers causes large documents to fail

[metmajer]

We use wkhtmltopdf to convert HTML to PDF. Using externalized headers and footers in dedicated HTML files causes wkhtmltopdf to open a file reference to each of these files for every page in the resulting PDF file -- and does not close these references until the entire PDF document has been fully generated.

Consequently, this causes issues with larger documents, as we unavoidably either produce a buffer overflow, or reach a system-wide limit for the maximum number of open files (ref. ulimit -n ). See wkhtmltopdf/wkhtmltopdf#2093 for more information. For larger documents ( > 100 pages), this behavior has resulted in socket timeout exceptions and document generation times of > 20 minutes.

A scalable solution will need to replace externalized HTML header and footer with a simplified, textual representation so that each document produces only a single open file handle.