remote_syslog tails one or more log files and sends syslog messages to a remote central syslog server. It generates packets itself, ignoring the system syslog daemon, so its configuration doesn't affect system-wide logging.
Uses:
- Collecting logs from servers & daemons which don't natively support syslog
- When reconfiguring the system logger is less convenient than a purpose-built daemon (e.g., automated app deployments)
- Aggregating files not generated by daemons (e.g., package manager logs)
This code is tested with the hosted log management service Papertrail and should work for transmitting to any syslog server.
remote_syslog2 is a rewrite of the ruby remote_syslog package. Not all features of the ruby version are supported, and there are some backwards incompatible changes.
Use remote_syslog2 (this README and application) unless you have a specific reason to use remote_syslog1.
- The syntax of some command-line arguments have changed slightly, though most are identical.
- Default hostname has been removed. Either the
hostconfig file option or the-dinvocation flag are required.
Precompiled binaries for Mac (Darwin), Linux and Windows are available on the remote_syslog2 releases page.
Untar the package, copy the "remote_syslog" executable into your $PATH, and then customize the included example_config.yml with the log file paths to read and the host/port to log to.
Optionally, move and rename the configuration file to /etc/log_files.yml so
that remote_syslog picks it up automatically. For example:
sudo cp ./remote_syslog /usr/local/bin
sudo cp example_config.yml /etc/log_files.yml
sudo vi /etc/log_files.yml
Configuration directives can also be specified as command-line arguments (below).
Usage of remote_syslog2:
-c, --configfile string Path to config (default "/etc/log_files.yml")
--debug-log-cfg string The debug log file; overridden by -D/--no-detach
-d, --dest-host string Destination syslog hostname or IP
-p, --dest-port int Destination syslog port (default 514)
-t, --dest-token string Destination ingestion token
--eventmachine-tail No action, provided for backwards compatibility
-f, --facility string Facility (default "user")
-h, --help Display this help message
--hostname string Local hostname to send from (default: OS hostname)
--log string Set loggo config, like: --log="<root>=DEBUG" (default "<root>=INFO")
--new-file-check-interval int How often to check for new files (seconds) (default 10)
-D, --no-detach Don't daemonize and detach from the terminal; overrides --debug-log-cfg
--no-eventmachine-tail No action, provided for backwards compatibility
--pid-file string Location of the PID file
--poll Detect changes by polling instead of inotify
-s, --severity string Severity (default "notice")
--tcp Connect via TCP (no TLS)
--tls Connect via TCP with TLS
-V, --version Display version and exit
Daemonize and collect messages from files listed in ./example_config.yml as
well as the file /var/log/mysqld.log. Write PID to /tmp/remote_syslog.pid
and send to port logs.papertrailapp.com:12345:
$ remote_syslog -c example_config.yml -p 12345 --pid-file=/tmp/remote_syslog.pid /var/log/mysqld.log
Stay attached to the terminal, look for and use /etc/log_files.yml if it
exists, and send with facility local0 to a.example.com:514:
$ remote_syslog -D -d a.example.com -f local0 /var/log/mysqld.log
Sample init files can be found in the examples directory. You may be able to:
$ cp examples/remote_syslog.init.d /etc/init.d/remote_syslog
$ chmod 755 /etc/init.d/remote_syslog
And then ensure it's started at boot, either by using:
$ sudo update-rc.d remote_syslog defaults
or by creating a link manually:
$ sudo ln -s /etc/init.d/remote_syslog /etc/rc3.d/S30remote_syslog
remote_syslog will daemonize by default.
Additional information about init files (init.d, supervisor, systemd and upstart) are
available in the examples directory.
If the receiving system supports sending syslog over TCP with TLS, you can
pass the --tls option when running remote_syslog:
$ remote_syslog -D --tls -p 1234 /var/log/mysqld.log
or add protocol: tls to your configuration file.
By default, remote_syslog looks for a configuration in /etc/log_files.yml.
The archive comes with a sample config. Optionally:
$ cp example_config.yml.example /etc/log_files.yml
log_files.yml has filenames to log from (as an array) and hostname and port
to log to (as a hash). Wildcards are supported using * and standard shell
globbing. Filenames given on the command line are additive to those in
the config file.
Only 1 destination server is supported; the command-line argument wins.
files:
- /var/log/httpd/access_log
- /var/log/httpd/error_log
- /var/log/mysqld.log
- /var/run/mysqld/mysqld-slow.log
destination:
host: logs.papertrailapp.com
port: 12345
protocol: tls
remote_syslog sends the name of the file without a path ("mysqld.log") as the syslog tag (program name).
After changing the configuration file, restart remote_syslog using the
init script or by manually killing and restarting the process. For example:
/etc/init.d/remote_syslog restart
Here's an advanced config which uses all options.
Provide --hostname somehostname or use the hostname configuration option:
hostname: somehostname
remote_syslog automatically detects and activates new log files that match
its file specifiers. For example, *.log may be provided as a file specifier,
and remote_syslog will detect a some.log file created after it was started.
By default, globs are re-checked every 10 seconds. To check for new files more
frequently, use the --new-file-check-interval argument. For example, to
recheck globs every 1 second, use:
--new-file-check-interval 1
Note: messages may be written to new files in the period between when the file is created and when the periodic glob check detects it. This data is not transmitted.
If globs are specified on the command-line, enclose each one in single-quotes
('*.log') so the shell passes the raw glob string to remote_syslog (rather
than the current set of matches). This is not necessary for globs defined in
the config file.
External log rotation scripts often move or remove an existing log file
and replace it with a new one (at a new inode). The Linux standard script
logrotate supports a copytruncate config
option. With that option, logrotate will copy files, operate on the copies,
and truncate the original so that the inode remains the same.
remote_syslog will handle both approaches seamlessly, so it should be no
concern as to which method is used. If a log file is moved or renamed,
and a new file is created (at a new inode), remote_syslog will follow that
new file at the new inode (assuming it has the same absolute path name). If
a file is copied then truncated, remote_syslog will seek to the beginning of
the truncated file and continue to read it.
Some logging programs such as Java's gclog (-XX:+PrintGC or -verbose:gc)
do not log in append mode, so if another program such as logrotate (set to
copytruncate) truncates the file, on the next write of the Java logger, the
OS will fill the file with NUL bytes upto the current offset of the file descriptor.
More info on that here.
remote_syslog will detect those leading NUL bytes, discard them, and log the discard count.
Provide one or more regular expressions to prevent certain files from being matched.
exclude_files:
- \.\d$
- .bz2
- .gz
There may be certain log messages that you do not want to be sent. These may be repetitive log lines that are "noise" that you might not be able to filter out easily from the respective application. To filter these lines, use the exclude_patterns with an array or regexes:
exclude_patterns:
- exclude this
- \d+ things
Run multiple instances to specify unique syslog hostnames.
To do that, provide an alternate PID path as a command-line option to the additional instance(s). For example:
--pid-file=/var/run/remote_syslog_2.pid
Note: Daemonized programs use PID files to identify whether the program is already running (more). Like other daemons, remote_syslog will refuse to run as a daemon (the default mode) when a PID file is present. If a .pid file is present but the daemon is not actually running, remove the PID file.
remote_syslog uses the log file name (like "access_log") as the syslog program name, or what the syslog RFCs call the "tag." This is ideal unless remote_syslog watches many files that have the same name.
In that case, tell remote_syslog to set another program name using the
tag attribute in the configuration file:
files:
- path: /var/log/httpd/access_log
tag: apache
destination:
host: logs.papertrailapp.com
port: 12345
protocol: tls
... or on the command line:
remote_syslog apache=/var/log/httpd/access_log
This functionality was introduced in version 0.17
To output debugging events with maximum verbosity, run:
remote_syslog --debug-log-cfg=logfile.txt --log="<root>=DEBUG"
.. as well as any other arguments which are used in normal operation. This
will set loggo's
root logger to the DEBUG level and output to logfile.txt.
To send messages longer than 1024 characters, use TCP (either TLS or cleartext TCP) instead of UDP. See "Sending messages securely" to use TCP with TLS for messages of any length.
Here's why longer UDP messages are impossible to send over the Internet.
When running remote_syslog in the foreground using the -D switch, if you
receive the error:
Error creating fsnotify watcher: inotify_init: too many open files
determine the maximum number of inotify instances that can be created using:
cat /proc/sys/fs/inotify/max_user_instances
and then increase this limit using:
echo VALUE >> /proc/sys/fs/inotify/max_user_instances
where VALUE is greater than the present setting. Confirm that remote_syslog starts
up and then apply this new value permanently by adding the following to
/etc/sysctl.conf::
fs.inotify.max_user_instances = VALUE
When monitoring a large number of files, this error may occur:
FATAL -- Error watching /path/here : no space left on device
To solve this, determine the maximum number of user watches that can be created using:
cat /proc/sys/fs/inotify/max_user_watches
and then increase them using:
echo VALUE >> /proc/sys/fs/inotify/max_user_watches
Once again, confirm that remote_syslog starts and then apply this value permanently by adding the following to /etc/sysctl.conf::
fs.inotify.max_user_watches = VALUE
- See whether the issue has already been reported: https://github.com/papertrail/remote_syslog2/issues/
- If you don't find one, create an issue with a repro case.
remote_syslog2 is written in go. To get everything set up, install go then run:
go install github.com/mitchellh/gox
git clone git@github.com:papertrail/remote_syslog2.git
To run tests:
# run all tests
go test ./...
# run all tests except the slower syslog reconnection tests
go test -short ./...
make
As of 0.18, we introduced ARM support for remote_syslog2. Current ARM builds support all ARM platforms with hardware floating point instruction sets. This includes All Raspberry PI devices, most ARMv6 chips (Cortex), and ARMv7 and beyond.
Once you've made your great commits: