TO: 638...@bugs.debian.org
And... Still a bug. Since at least 2010...
As of December 2022 this is still an issue in 3.1.23.1-1
ii at 3.1.23-1.1 arm64 Delayed job execution and
batch processing
ls -l /usr/sbin/atd
-rwxr-xr-x 1 root root 26568 Sep 25 2020 /usr/sbin/atd
/var/log/syslog (and/or /var/log/cron) filled with:
...
Dec 5 21:11:36 mysys atd[17889]: File a006e001a8bf99 is in wrong format
- aborting
Dec 5 21:11:36 mysys atd[17890]: File a006e001a8bf99 is in wrong format
- aborting
Dec 5 21:11:36 mysys atd[17891]: File a006e001a8bf99 is in wrong format
- aborting
Dec 5 21:11:36 mysys atd[17892]: File a006e001a8bf99 is in wrong format
- aborting
...
/var/spool/cron/atjobs contained:
drwxrwx--T 2 daemon daemon 4096 Dec 5 21:11 .
drwxr-xr-x 5 root root 4096 Dec 4 11:30 ..
-rw------- 1 daemon daemon 6 Dec 4 13:18 .SEQ
-rwx------ 2 dab daemon 0 Dec 4 13:18 '=006e001a8bf99'
-rwx------ 2 dab daemon 0 Dec 4 13:18 a006e001a8bf99
Deleting these zero-sized files/jobs and restarting atd resolved the issue.
I'm not a developer, but a possible quick fix would be to add:
find /var/spool/cron/atjobs/ -type f -empty -print -delete
to the /etc/init.d/atd start stanza
Although this would only cover instances when/where atd is (re)started.