We're running a Redhat server ( and experienced some odd behaviour with our cron setup. We're running an out-of-the-box installation of cron.

We have a task that runs daily at 2am and has done so successfully for many months. However yesterday (27/03/2011) morning 8 instances of the task were started.

Here is the relevant section of our cron log file:

Mar 27 02:00:01 fs1 CROND[19557]: (root) CMD (true && /home/youdev/you_server/NEW_FS1/housekeeping_scripts/copy_pdf_printer_files_to_fs2.sh)
Mar 27 02:00:01 fs1 CROND[19563]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:00:02 fs1 CROND[19597]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:00:12 fs1 CROND[19931]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:00:22 fs1 CROND[20060]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:00:32 fs1 CROND[20067]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:00:42 fs1 CROND[20077]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:00:52 fs1 CROND[20084]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:01:02 fs1 CROND[20094]: (youdev) CMD (true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh)
Mar 27 02:01:02 fs1 CROND[20095]: (root) CMD (run-parts /etc/cron.hourly)
Mar 27 02:01:02 fs1 CROND[20096]: (root) CMD (true && /home/youdev/you_server/NEW_FS1/housekeeping_scripts/copy_pdf_printer_files_to_fs2.sh)
Mar 27 02:01:02 fs1 run-parts(/etc/cron.hourly)[20095]: starting 0anacron

Each of the fetch_db_data.sh scripts started correctly (they're just a shell script that calls a rsync - and the rsync tmp files exist as do the files that were successfully rsynced across).

This is the only job we have that starts at 02:00, yet we have a job that runs on the minute (copy_pdf_printer_files_to_fs2.sh) and this correctly ran just once at 02:00.

The only thing that we can think could have caused it is a daylight saving issue (we're in the UK so went from GMT to BST at 02:00 on the day in question) - however all the digging we've done says that cron handles daylight saving issues fine. And even then, shouldn't we have expected 0 runs (as time skipped from 00:59 to 02:00) rather than 8?

Here is the crontab entry:


# BACKUP LATEST DB FROM LIVE (takes approx one hour)
0 2 * * * true && /home/youdev/you_server/NEW_FS1/backup_scripts/fetch_db_data.sh

Many thanks in advance for reading this and for any help and suggestions,


  • Without making any other comment, I note that on Sunday morning the clock skipped from 0100 to 0200, not 0300, so you would still have expected it to run. We only went forward one hour, not two!
    – MadHatter
    Mar 28, 2011 at 15:04
  • Yes, but with respect, you edited it wrong. The time "skipped" from 00:59:59GMT to 02:00:00BST, not 01:59:59GMT to 03:00:00BST; we really did have an 0200 on Sunday morning. The Wikipedia article on British Summer Time references the Summer Time Order 2002 as saying that BST applies during "the period beginning at one o'clock, Greenwich mean time, in the morning of the last Sunday in March and ending at one o'clock, Greenwich mean time, in the morning of the last Sunday in October"
    – MadHatter
    Mar 28, 2011 at 15:48

2 Answers 2


I'd use Ockham's Razor with this. It's bound to be the Summer Time clock changes.

  • 2
    You're more than likely right - but why would it trigger 8 ? I could see the case for 1 extra (maybe even a couple more than that) but 8 seems really quite odd.
    – ChrisW
    Mar 28, 2011 at 15:42
  • 1
    I've had a couple of Servers repeat cron jobs around the start of the hour but never eight times ! Once was down to a manual cronned NTP entry jumping the clock back a few seconds each hour. Are you running NTP ? Mar 28, 2011 at 15:45
  • Jonathan's answer is probably more likely than it being DST changes IMHO. In my experience DST could account for 2 runs, but not 8. Note that you can avoid all of this trouble by setting the server's clock to UTC (Just Sayin' -- I don't do it myself :P)
    – voretaq7
    Mar 28, 2011 at 16:01
  • Yes, ntpd is running (and was running at the time)
    – ChrisW
    Mar 28, 2011 at 16:08
  • I think maybe a combination of NTP and DST caused the bizarre number of Cron runs. Ockham's Razor again :-) Mar 29, 2011 at 6:17

Please see Daniël van Eeden answer to question Cron job executed multiple times daylight saving time:

It is a known bug:



I've seen it happen on CentOS 6.2 with cronie-1.4.4-7.el6.x86_64 on the 31st of march (CET -> CEST)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.