Search

Top 60 Oracle Blogs

Recent comments

Getting rid of annoying, repetitive messages in /var/log/messages

The primary source of information regarding any change or issue on a linux system is the /var/log/messages file. I am often annoyed when a linux system is setup in such a way that certain messages are written to syslog with a high frequency swamping the messages file with information that is not important. The reason for my annoyance is that this makes it very hard to actually spot important information because you have to skip through a lot of lines before you find the important information, especially if you do not know for sure if there a message in the first place.

Please mind this blogpost is created on a Centos 7 server which uses rsyslog.

There are a couple of ways to manage this. The standard syslog way of managing this is the following, which can be found in /etc/rsyslog.conf:

#### RULES ####

# Log all kernel messages to the console.
# Logging much else clutters up the screen.
#kern.*                                                 /dev/console

# Log anything (except mail) of level info or higher.
# Don't log private authentication messages!
*.info;mail.none;authpriv.none;cron.none                /var/log/messages

# The authpriv file has restricted access.
authpriv.*                                              /var/log/secure

# Log all the mail messages in one place.
mail.*                                                  -/var/log/maillog


# Log cron stuff
cron.*                                                  /var/log/cron

# Everybody gets emergency messages
*.emerg                                                 :omusrmsg:*

# Save news errors of level crit and higher in a special file.
uucp,news.crit                                          /var/log/spooler

# Save boot messages also to boot.log
local7.*                                                /var/log/boot.log

What this shows is that certain principal important tasks have their own ‘facility’, which in this example are ‘mail’, ‘cron’, ‘authpriv’, etc. shown above, which are put in their own file.

But how about processes that do not have their own facility, and just write to syslog? One example of these this is dhclient, which produces messages like these:

Jun 30 03:28:32 ip-172-31-12-40 dhclient[19604]: DHCPREQUEST on eth0 to 172.31.0.1 port 67 (xid=0x5dfd88c4)
Jun 30 03:28:32 ip-172-31-12-40 dhclient[19604]: DHCPACK from 172.31.0.1 (xid=0x5dfd88c4)
Jun 30 03:28:35 ip-172-31-12-40 dhclient[19604]: bound to 172.31.12.40 -- renewal in 1726 seconds.
Jun 30 03:57:21 ip-172-31-12-40 dhclient[19604]: DHCPREQUEST on eth0 to 172.31.0.1 port 67 (xid=0x5dfd88c4)
Jun 30 03:57:21 ip-172-31-12-40 dhclient[19604]: DHCPACK from 172.31.0.1 (xid=0x5dfd88c4)
Jun 30 03:57:23 ip-172-31-12-40 dhclient[19604]: bound to 172.31.12.40 -- renewal in 1798 seconds.

Which it does every 5 minutes. This is actually truly annoying…

Luckily, there is a solution: rsyslog has the option to filter based on more properties than the logging facility a process is using. This is done using a script in /etc/rsyslog.d/.

On my server, the majority of the messages are from the dhclient and systemd daemons, and both the messages seem to be informal. In order not to miss anything, I still want that information to be logged, but not in the /var/log/messages file.

This can be actually quite simply be accomplished using the following two scripts in /etc/rsyslog.d/:

dhclient.conf:

if $programname == 'dhclient' then /var/log/dhclient.log
&stop

systemd.conf

if $programname = 'systemd' then /var/log/systemd.log
&stop

Once you created these scripts, you need to make rsyslogd read this new configuration. I thought killall -HUP rsyslogd would accomplish this, but outside of a message in the /var/log/messages file saying it got the HUP signal, it doesn’t execute a new task.

However, executing:

systemctl stop rsyslog.service
systemctl start rsyslog.service

Does make rsyslog read the new configuration, and then both dhclient and systemd log to their own files and do not write to the messages file anymore!

There is one last thing that needs to be done: make sure the newly defined logfiles are cleaned just like all the other files in /var/log. Otherwise these files will endlessly grow, eventually occupying all the space in the filesystem where /var/log is part of.

This too is really easy, the newly defined logfiles can be added to the list of syslog files for logrotate, which is defined in /etc/logrotate.d/syslog:

/var/log/cron
/var/log/maillog
/var/log/messages
/var/log/secure
/var/log/spooler
/var/log/dhclient.log
/var/log/systemd.log
{
    missingok
    sharedscripts
    postrotate
	/bin/kill -HUP `cat /var/run/syslogd.pid 2> /dev/null` 2> /dev/null || true
    endscript
}

As you can see, I simply added /var/log/dhclient.log and /var/log/systemd.log to the list.

Please mind that the filter for dhclient and systemd is the executable name, so even if the severity of the logging message is high, a message from these daemons will still go to the file it is configured to log to.