Recently, probably due to an unexpected power failure, the man page cache on my machine which is managed by mandb got corrupt. I noticed when systemctl showed the man-db.service as failed:
% sudo systemctl ... ● man-db.service loaded failed failed Update man-db cache ...
systemctl status didn't reveal that much information on the issue:
% sudo systemctl status man-db.service ● man-db.service - Update man-db cache Loaded: loaded (/usr/lib/systemd/system/man-db.service; static; vendor preset: disabled) Active: failed (Result: exit-code) since Sam 2016-11-26 16:47:53 CET; 10min ago Process: 23481 ExecStart=/usr/bin/mandb (code=exited, status=1/FAILURE) Process: 23475 ExecStart=/usr/bin/mkdir -p /var/cache/man (code=exited, status=0/SUCCESS) Main PID: 23481 (code=exited, status=1/FAILURE) Nov 26 16:47:53 tuxedo mandb: Handbuchseiten unter /usr/share/man/ru.KOI8-R werden verar Nov 26 16:47:53 tuxedo mandb: Alte Datenbankeinträge in /usr/share/man/tr werden gelösch Nov 26 16:47:53 tuxedo mandb: Handbuchseiten unter /usr/share/man/tr werden verarbeitet Nov 26 16:47:53 tuxedo mandb: Alte Datenbankeinträge in /usr/share/man/fi werden gelösch Nov 26 16:47:53 tuxedo mandb: Handbuchseiten unter /usr/share/man/fi werden verarbeitet Nov 26 16:47:53 tuxedo mandb: Alte Datenbankeinträge in /usr/share/man/zh werden gelösch Nov 26 16:47:53 tuxedo systemd: man-db.service: Main process exited, code=exited, status=1/F Nov 26 16:47:53 tuxedo systemd: Failed to start Update man-db cache. Nov 26 16:47:53 tuxedo systemd: man-db.service: Unit entered failed state. Nov 26 16:47:53 tuxedo systemd: man-db.service: Failed with result 'exit-code'.
So I ran mandb manually to figure out what was wrong:
% sudo mandb Purging old database entries in /usr/share/man... Processing manual pages under /usr/share/man... Purging old database entries in /usr/share/man/sr@latin... Processing manual pages under /usr/share/man/sr@latin... ... Purging old database entries in /usr/share/man/zh... gdbm fatal: Unexpected end of file
So gdbm seemed to have some problem. Since I didn't know what gdbm was, I asked the all knowingly package manager of ArchLinux:
% pacman -Qi gdbm Name : gdbm Version : 1.12-2 Description : GNU database library ...
Since it is a database library I assumed that the database must be corrupt. Lucky enough mandb provides an option to create the database from scratch:
% sudo mandb -c Processing manual pages under /usr/share/man... Updating index cache for path `/usr/share/man/man1'. Wait...mandb: warning: /usr/share/man/man1/gdk-pixbuf-csource.1.gz: whatis parse for gdk-pixbuf-csource(1) failed mandb: warning: /usr/share/man/man1/gdk-pixbuf-query-loaders.1.gz: whatis parse for gdk-pixbuf-query-loaders(1) failed ... Processing manual pages under /usr/local/man... 124 man subdirectories contained newer manual pages. 22844 manual pages were added. 0 stray cats were added.
It spilled my screen with lots of output and warnings, but seemed to finish properly.
It didn't help though and mandb still failed. Looking at the output again I noticed the line Purging old database entries in /usr/share/man/zh.... To see which package placed its manpage there I queried pacman again:
% LANG=C pacman -Qo /usr/share/man/zh /usr/share/man/zh/ is owned by nmap 7.31-1
So I tried to uninstall and reinstall nmap:
% sudo pacman -R nmap % sudo mandb % sudo pacman -S nmap % sudo mandb
And this time it finally worked! I'm still not sure what exactly triggered the failure in the first place, but I'm not sure how to further debug it.