On Sun, 26 Feb 2023 19:39:51 +0100 "Carlos E. R." <robin.listas@telefonica.net> wrote:
On 2023-02-26 19:30, Felix Miata wrote:
Carlos E. R. composed on 2023-02-26 09:25 (UTC+0100):
Felix Miata wrote:
# journalctl -b --no-hostname| grep "Feb 25" | egrep -v 'smartd|rpc' Feb 25 00:00:47 systemd[1]: Starting Do daily mandb update... Feb 25 00:00:47 systemd[1]: Starting Rotate log files... Feb 25 00:00:47 systemd[1]: Starting Update locate database... Feb 25 00:00:47 systemd[1]: logrotate.service: Deactivated successfully. Feb 25 00:00:47 systemd[1]: Finished Rotate log files. Feb 25 00:00:47 systemd[1]: mandb.service: Deactivated successfully. Feb 25 00:00:47 systemd[1]: Finished Do daily mandb update.
What is mystifiying to me is that the three tasks start simultaneously, and the three end in zero seconds.
Why have multiple cores dividing up one job at a time when several jobs can each run on one core without need to manage separation and recombination?
Sure, the CPU part can be divided. But these jobs are about shufling files or seeking files.
Err, the mandb is probably a NO-OP unless a new program has been installed, and rotate logs is just close some files and open new ones.
The updatedb job in particular has to explore every directory in every hard disk, meaning a log of disk head movement, impeding any other hard disk job and impacting the entire machine down. There is no parallelization in hard disk.
As you say updatedb has to do more work and that is exactly the one that started but did not finish, so I'm still not sure what your concern is. It feels like you're worrying about nothing.