Index: [Article Count Order] [Thread]

Date:  Sat, 02 Feb 2008 16:15:00 +0000
From:  Dogsbody <dan (at mark) dogsbody.org>
Subject:  [coba-e:11880] Re: run-parts /etc/cron.hourly
To:  coba-e (at mark) bluequartz.org
Message-Id:  <47A49704.1060108 (at mark) dogsbody.org>
In-Reply-To:  <20080201233308.M98297 (at mark) domain-nameit.net>
References:  <20080201233308.M98297 (at mark) domain-nameit.net>
X-Mail-Count: 11880


> I've been getting lots of these messages.
> Looks like the entire system has outgrown itself.

Not really, if you look at the errors you will see "Trying other mirror" so yum 
has realised there is a problem and moved onto another mirror were it got the 
correct data.

It's only the NuOnce and the BlueQuartz repo which are single servers and so are 
a single point of failure although it's not the end of the world if they go down 
it does mean errors like this.

Last week I suggested how people can help the BQ project by running a mirror of 
the BQ repository however no one seems to be interested, at least no one has 
replied saying they have set up a mirror which is a shame considdering the 
number of ISP's offering BQ.

> /etc/cron.hourly/yum_checker:

I personally feel that this *hourly* script is overkill and delete it from all 
my servers...
$ rm /etc/cron.hourly/yum_checker
$ rm /tmp/yum.check-update


I replace it with a *daily* check that mails me if anything needs doing...
$ touch /etc/cron.daily/yum-check
$ chmod 755 /etc/cron.daily/yum-check
$ vi /etc/cron.daily/yum-check
#!/bin/sh
SERVER=`hostname`
OUTPUT=`/usr/bin/yum check-update`
EXITCODE=$?
if [ ${EXITCODE} != 0 ]
then
     echo "${OUTPUT}" | /bin/mail -s "${SERVER}: Yum Updater" root
fi


If I get a mail then I can manually go in and do a `yum update`, that way I can 
check that if XYZ application was updated it was properly restarted and still 
working fine.  I realise this may not work for other people, just sharing what I 
do :-)

Dan