Logs

Each Tarantool instance logs important events to its own log file. For instances started with tt, the log location is defined by the log_dir parameter in the tt configuration. By default, it's /var/log/tarantool in the tt system mode, and the var/log subdirectory of the tt working directory in the local mode. In the specified location, tt creates separate directories for each instance's logs.

To check how logging works, write something to the log using the log module:

 
$ tt connect application   • Connecting to the instance...   • Connected to applicationapplication> require('log').info("Hello for the manual readers")---...

Then check the logs:

 
$ tail instances.enabled/application/var/log/instance001/tt.log2024-04-09 17:34:29.489 [49502] main/106/gc I> wal/engine cleanup is resumed2024-04-09 17:34:29.489 [49502] main/104/interactive/box.load_cfg I> set 'instance_name' configuration option to "instance001"2024-04-09 17:34:29.489 [49502] main/104/interactive/box.load_cfg I> set 'custom_proc_title' configuration option to "tarantool - instance001"2024-04-09 17:34:29.489 [49502] main/104/interactive/box.load_cfg I> set 'log_nonblock' configuration option to false2024-04-09 17:34:29.489 [49502] main/104/interactive/box.load_cfg I> set 'replicaset_name' configuration option to "replicaset001"2024-04-09 17:34:29.489 [49502] main/104/interactive/box.load_cfg I> set 'listen' configuration option to [{"uri":"127.0.0.1:3301"}]2024-04-09 17:34:29.489 [49502] main/107/checkpoint_daemon I> scheduled next checkpoint for Tue Apr  9 19:08:04 20242024-04-09 17:34:29.489 [49502] main/104/interactive/box.load_cfg I> set 'metrics' configuration option to {"labels":{"alias":"instance001"},"include":["all"],"exclude":}2024-04-09 17:34:29.489 [49502] main I> entering the event loop2024-04-09 17:34:38.905 [49502] main/116/console/unix/:/tarantool I> Hello for the manual readers

When logging to a file, the system administrator must ensure logs are rotated timely and do not take up all the available disk space. The recommended way to prevent log files from growing infinitely is using an external log rotation program, for example, logrotate, which is pre-installed on most mainstream Linux distributions.

A Tarantool log rotation configuration for logrotate can look like this:

 
# /var/log/tarantool/<env>/<app>/<instance>/*.log/var/log/tarantool/*/*/*/*.log {    daily    size 512k    missingok    rotate 10    compress    delaycompress    sharedscripts # Run tt logrotate only once after all logs are rotated.    postrotate        /usr/bin/tt -S logrotate    endscript}

In this configuration, tt logrotate is called after each log rotation to reopen the instance log files after they are moved by the logrotate program.

There is also the built-in function log.rotate(), which you can call on an instance to reopen its log file after rotation.

Tarantool can write its logs to a log file, to syslog, or to a specified program through a pipe. For example, to send logs to syslog, specify the log.to parameter as follows:

 
log:  to: syslog  syslog:    server: '127.0.0.1:514'