Sometime it happen that the command df and the command du report different results such as:
df -h /tmp Filesystem Size Used Avail Use% Mounted on /dev/vzfs 16G 12G 4G 75% /tmp |
While a du command show somethign different such as:
du -hs /tmp/ 10G /tmp/ |
There are usually different reason when this happen.
Continue reading »
Logging is a critical thing for all system administrators, if you log too much and you don’t manage the files you could fill up a partition or even worst stop some service, if you don’t log enough you’ll lose information when something goes wrong, in general a good solution for this is to send all the logs to a central server that will store for the time you need them, and keep just 1,2 days of log into the local machine.
You could do this configuration easily with rsyslog or syslog-ng to send/receive the logs and logrotate to rotate the files locally on your machines, today I want to show you some open source programs that can receive the logs, store them on filesystem or database and analyse them presenting the results via Web dashboards.
These are large applications most suited for big company, or in general to everyone that want to keep and manage a lot of data they are: Apache Flume, Logstash, Greylog2 and Scribe
Varnish is a great tool that accelerates the performance of your site with a simple configuration and a low usage of resources, compared to the result you’ll get. Just because Varnish is designed and engineered to be fast by default it does not write any log to disk, outrageous you could think … to be honest the logs are available in a memory segment when this memory is filled, Varnish starts from the beginning and overwrites the oldest data in an infinite loop.
So in reality there are logs and they are available for some time, and this solution is much, much faster then logging to a file and it doesn’t require disk space, but how we can get them on file, possibly divided by virtual host as do Apache or Nginx ?
Continue reading »