RDSLogs

rdslogs is a tool to download or stream log files from RDS. When streaming, you can choose to stream them to STDOUT or directly to Honeycomb.

To learn more about using Honeycomb, see our docs (and RDS-specific docs).

The default action of rdslogs is to stream the current log file. Use the --download flag to download log files instead.

The default output is STDOUT to see what's happening:

rdslogs --region us-east-1 --identifier my-rds-database

To output the results directly to Honeycomb, use the --output honeycomb flag and include the --writekey and --dataset flags. Optionally, the --sample_rate flag will only send a portion of your traffic to Honeycomb.

rdslogs --region us-east-1 --identifier my-rds-database --output honeycomb --writekey abcabc123123 --dataset "rds logs"

Deprecation Notice for MySQL, MariaDB, and Aurora

rdslogs is deprecated for MySQL, MariaDB, and Aurora: please use Cloudwatch Logs combined with our Agentless Integrations for AWS.

rdslogs relies on the RDS API to tail mysql logs in realtime. Due to a bug in the API, slow query logs can randomly "disappear" for long periods of time, leaving large gaps in your MySQL Dataset. Amazon has acknowledged the bug, but has no ETA, so we have deprecated this tool in favor of Cloudwatch Logs, which are more reliable.

Installation

rdslogs is available as a .deb or .rpm package from honeycombio; see the MySQL RDS or PostgreSQL RDS integration documentation for links and command line instructions.

When installed from a package, there is a config file at /etc/rdslogs/rdslogs.conf. Instead of using the command line flags as indicated in the previous section, edit the config file with the intended values. After doing so, start the service with the standard sudo initctl start rdslogs (upstart) or sudo systemctl start rdslogs (systemd) commands.

To build and install directly from source:

go get github.com/honeycombio/rdslogs

Usage

$ rdslogs --help
Usage:
  rdslogs rdslogs --identifier my-rds-instance

rdslogs streams a log file from Amazon RDS and prints it to STDOUT or sends it
up to Honeycomb.io.

AWS Requirements

AWS credentials are required and can be provided via IAM roles, AWS shared config (~/.aws/config), AWS shared credentials (~/.aws/credentials), or the environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. Below is the minimal IAM policy needed by RDSLogs.

{
    "Version": "2012-10-17",
    "Statement": [
    {
        "Effect": "Allow",
        "Action": [
            "rds:DescribeDBInstances",
            "rds:DescribeDBLogFiles",
            "rds:DownloadDBLogFilePortion"
        ],
        "Resource": "*"
    }
  ]
}

Passing --download triggers Download Mode, in which rdslogs will download the specified logs to the directory specified by --download_dir. Logs are specified via the --log_file flag, which names an active log file as well as the past 24 hours of rotated logs. (For example, specifying --log_file=foo.log will download foo.log as well as foo.log.0, foo.log.2, ... foo.log.23.)

When --output is set to honeycomb, the --writekey and --dataset flags are required. Instead of being printed to STDOUT, database events from the log will be transmitted to Honeycomb. --scrub_query and --sample_rate also only apply to Honeycomb output.

Application Options:
      --region=               AWS region to use (default: us-east-1)
  -i, --identifier=           RDS instance identifier
      --dbtype=               RDS database type. Accepted values are mysql and postgresql.
                              (default: mysql)
      --log_type=             Log file type. Accepted values are query and audit. Audit is
                              currently only supported for mysql. (default: query)
  -f, --log_file=             RDS log file to retrieve
  -d, --download              Download old logs instead of tailing the current log
      --download_dir=         directory in to which log files are downloaded (default: ./)
      --num_lines=            number of lines to request at a time from AWS. Larger number will
                              be more efficient, smaller number will allow for longer lines
                              (default: 10000)
      --backoff_timer=        how many seconds to pause when rate limited by AWS. (default: 5)
  -o, --output=               output for the logs: stdout or honeycomb (default: stdout)
      --writekey=             Team write key, when output is honeycomb
      --dataset=              Name of the dataset, when output is honeycomb
      --api_host=             Hostname for the Honeycomb API server (default:
                              https://api.honeycomb.io/)
      --scrub_query           Replaces the query field with a one-way hash of the contents
      --sample_rate=          Only send 1 / N log lines (default: 1)
  -a, --add_field=            Extra fields to send in request, in the style of "field:value"
  -v, --version               Output the current version and exit
  -c, --config=               config file
      --write_default_config  Write a default config file to STDOUT
      --debug                 turn on debugging output

Help Options:
  -h, --help                  Show this help message