Sematext Syslog-ng Integration
Overview¶
syslog-ng is a modern syslog daemon that's focused on flexibility and portability. It also has an easy to use configuration format that helps you ship your logs to a Logs App in 3 steps:
- sources. syslog-ng can listen to local syslog traffic, can tail files and more
- destinations. You can send your logs to Logs Management App via UDP, TCP or RFC-5425 TLS Syslog
- bind sources to destinations. Once you have your sources and destinations defined, you have to define paths by linking sources and destinations with the log statement
Configure Sources¶
Sources enable syslog-ng collect the logs you want to send to Logs Management App. You can use system() to collect all the local syslog messages from that system. You can put this at the beginning of your /etc/syslog-ng/syslog-ng.conf, along with your configuration version. We recommend running version 3.19 or later:
@version:3.19
source local_logs {
internal();
system();
};
To download a more recent version of syslog-ng than the one on your system, take a look at the download page of syslog-ng.
Tailing Files¶
syslog-ng can also listen to other inputs, such as files, TCP or UDP.
To tail a file, you'll use the file() source and point it to the right file. If that file doesn't contain syslog-formatted messages, you'll need to tell syslog-ng not to try and parse it, and if you need multi-line support, you should specify that, too:
source jetty_log { file("/var/log/jetty" flags(no-parse) multi-line-mode(indented)); };
Configure Destinations¶
There are two options to properly configure your destination:
- with your Logs Management application token (which is what we recommend)
- or by authorizing your public IPs from the Logs Management App UI.
After you choose your authentication method, you'll configure your destination according to the protocol you prefer: UDP, TCP or TLS. For all cases, your endpoint will be logsene-syslog-receiver.sematext.com / logsene-syslog-receiver.eu.sematext.com (if using Sematext Cloud Europe)
Token Authentication¶
To enable token authentication, get the token of your Logs Management application from the list of Logs Maangement applications. Then, put it in the template() statement of your Logs Management App destination. See working examples below. The end result is a CEE-formatted JSON syslog message that contains your token in the logsene-app-token field.
IMPORTANT: For JSON formatting that enables token authentication to work you'll need syslog-ng version 3.3 or later and the libjson syslog-ng plugin. The libjson plugin is typically provided by the syslog-ng-json package of your distribution.
IP-based Authentication¶
The alternate method for authentication is by pre-authorizing your public IP in Logs Management App's UI. Here is a complete guide on how to do that. If you choose this path, you don't need the template() statement from the code snippets below.
UDP¶
To send logs over the fire-and-forget UDP protocol, your can add a destination like the one below to /etc/syslog-ng/syslog-ng.conf. If you're using token-based authentication, you'll have to fill in your Logs Management application token. If you've authorized your public IP, the template() statement should be removed:
template sematext {
template("<${PRI}>1 ${ISODATE} ${HOST} ${PROGRAM} ${PID} - - $(format-json --pair message=\"$MSG\" --pair logsene-app-token=\"LOGSENE_APP_TOKEN_GOES_HERE\")\n");
};
destination sematext {
udp("logsene-syslog-receiver.sematext.com"
port(514)
# template() statement should be removed if you authorized your IP
template(sematext)
);
};
TCP¶
For TCP, the destination will look similar to the one for UDP, the only change is in the protocol() statement. Make sure your real token is there or remove the template() statement if you've authorized your IP:
template sematext {
template("<${PRI}>1 ${ISODATE} ${HOST} ${PROGRAM} ${PID} - - $(format-json --pair message=\"$MSG\" --pair logsene-app-token=\"LOGSENE_APP_TOKEN_GOES_HERE\")\n");
};
destination sematext {
tcp("logsene-syslog-receiver.sematext.com"
port(514)
# template() statement should be removed if you authorized your IP
template(sematext)
);
};
TLS¶
For configuring RFC-5425 TLS Syslog, you'll configure the destination in a similar fashion to plain TCP, except for adding tls() statement and pointing it to your certificates directory and changing the port to 10514:
template sematext {
template("<${PRI}>1 ${ISODATE} ${HOST} ${PROGRAM} ${PID} - - $(format-json --pair message=\"$MSG\" --pair logsene-app-token=\"LOGSENE_APP_TOKEN_GOES_HERE\")\n");
};
destination sematext {
tcp("logsene-syslog-receiver.sematext.com"
port(10514)
tls(
ca_dir("/etc/ssl/certs")
peer_verify("required_trusted")
)
# template() statement should be removed if you authorized your IP
template(sematext)
);
};
Bind Sources to Destinations¶
After configuring your source and destination the last step is binding the two:
log {
source(local_logs); destination(sematext);
};
If you're tailing files or you defined other sources, you need to bind them to the logsene destination in order to have those messages shipped to your Logs Management application as well. For example:
log {
source(jetty_log); destination(sematext);
};
Then, restart syslog-ng and you should see your logs in the Logs Management App UI.
Logs Tagging¶
From your syslog messages, Logs Management App will populate a number of special fields, such as the source and host. You can also configure syslog-ng to add a tag to logs matching certain criteria. This is useful when you want to quickly identify a special kind of logs. For example, you could tag events that come to the "kern" facility with a severity/level of "err" as "kernel errors".
To achieve this, you can define a filter that matches such events. Then, you'd define a destination similar to the ones described above, where you'd add a tags field. Finally, you'd define a log statement where you bind your source, the newly defined filter and the newly defined destination:
filter user_tests { facility(kern) and level(err) };
template sematext {
template("<${PRI}>1 ${ISODATE} ${HOST} ${PROGRAM} ${PID} - - $(format-json --pair message=\"$MSG\" --pair logsene-app-token=\"LOGSENE_APP_TOKEN_GOES_HERE\")\n");
};
destination sematext {
tcp("logsene-syslog-receiver.sematext.com"
port(514)
template(sematext)
);
};
log { source(all_syslog); filter(user_tests); destination(sematext); flags(final); };
# main Logsene "log" statement will be defined below
Notice the final flag to this log statement - this prevents syslog-ng from sending matched events twice (once with tags and once without). Make sure you place the log statement with tags before your main log statement.