Virus update times

shrek-m at shrek-m at GMX.DE
Sun Feb 29 20:52:20 GMT 2004

Kevin Spicer wrote:

>Theres been some discussion on the clamav list recently about the
>frequency of clients pulling database updates from their servers.  the
>most notable point was that several of the clam developers urged users
>to schedule their cron jobs at a random minute past the hour to try and
>get a better distribution of load on the servers.  I guess similar
>problems also afflict users of commercial scanners.
>Several things stuck me.
>1) Many (most?) MailScanner users use cron.hourly to schedule updates,
>therefore we, as a community, are probably responsible for a
>substantially increased load at one point every hour.
>2) Everyone updating at the same time increases the possibility of
>individual updates failing due to bandwidth/ server issues
>3) Any problems with the virus database introduced immediately before
>the point we all update are likely to affect all of us before they get
>4) We all have the same window of opportunity in our update cycles
>during which a new virus could propagate very quickly, at least if we
>all updated at different times we may stand a better chance of slowing
>the rate of spread.
>I therefore propose that update_virus_scanners be moved from
>/etc/cron.hourly to a file in /etc/cron.d and that the minute at which
>it is scheduled in that file be generated either at random or be the
>same as the minute at which the file was installed.  Obviously this
>would involve generating the file as part of the install process.

could it be possible to set this in update_virus_scanners with a random
i hope that this would not stop other scripts in cron.hourly.

# vi /usr/sbin/update_virus_scanners


sleep 300


# crontab -e -umailscanner-user


More information about the MailScanner mailing list