awslogs is a simple command line tool for querying groups, streams and events from Amazon CloudWatch logs.

One of the most powerful features is to query events from several streams and consume them (ordered) in pseudo-realtime using your favourite tools such as grep:

$ awslogs get /var/log/syslog ip-10-1.* --start='2h ago' | grep ERROR

• Aggregate logs from across streams.
  • Aggregate all streams in a group.
  • Aggregate streams matching a regular expression.
• Colored output.
• List existing groups
  • $ awslogs groups
• List existing streams
  • $ awslogs streams /var/log/syslog
• Watch logs as they are created
  • $ awslogs get /var/log/syslog ALL --watch
• Human-friendly time filtering:
  • --start='23/1/2015 14:23'
  • --start='2h ago'
  • --start='2d ago'
  • --start='2w ago'
  • --start='2d ago' --end='1h ago'
• Retrieve event metadata:
  • --timestamp Prints the creation timestamp of each event.
  • --ingestion-time Prints the ingestion time of each event.

Running: awslogs get /var/logs/syslog ALL -s1d will return you events from any stream in the /var/logs/syslog group generated in the last day.

You can easily install awslogs using pip:

$ pip install awslogs

If you are on OSX El Capitan, use the following (Why? Check Donald Stufft's comment here)

$ pip install awslogs --ignore-installed six

You can also install it with brew:

$ brew install awslogs

• awslogs groups: List existing groups
• awslogs streams GROUP: List existing streams withing GROUP
• awslogs get GROUP [STREAM_EXPRESSION]: Get logs matching STREAM_EXPRESSION in GROUP.
  • Expressions can be regular expressions or the wildcard ALL if you want any and don't want to type .*.

Note: You need to provide to all these options a valid AWS region using --aws-region or AWS_REGION env variable.

While querying for logs you can filter events by --start -s and --end -e date.

• By minute:

  • --start='2m' Events generated two minutes ago.
  • --start='1 minute' Events generated one minute ago.
  • --start='5 minutes' Events generated five minutes ago.

• By hours:

  • --start='2h' Events generated two hours ago.
  • --start='1 hour' Events generated one hour ago.
  • --start='5 hours' Events generated five hours ago.

• By days:

  • --start='2d' Events generated two days ago.
  • --start='1 day' Events generated one day ago.
  • --start='5 days' Events generated five days ago.

• By weeks:

  • --start='2w' Events generated two week ago.
  • --start='1 week' Events generated one weeks ago.
  • --start='5 weeks' Events generated five week ago.

• Using specific dates:

  • --start='23/1/2015 12:00' Events generated after midday on the 23th of January 2015.
  • --start='1/1/2015' Events generated after midnight on the 1st of January 2015.
  • --start='Sat Oct 11 17:13:46 UTC 2003' You can use detailed dates too.

  Note, for time parsing awslogs uses dateutil.

• All previous examples are applicable for --end -e too.

You can use --filter-pattern if you want to only retrieve logs which match one CloudWatch Logs Filter pattern. This is helpful if you know precisely what you are looking for, and don't want to download the entire stream.

For example, if you only want to download only the report events from a Lambda stream you can run:

$ awslogs get my_lambda_group --filter-pattern="[r=REPORT,...]"

Full documentation of how to write patterns: https://docs.aws.amazon.com/AmazonCloudWatch/latest/DeveloperGuide/FilterAndPatternSyntax.html

In a similar way than the aws-cli command, you can use --query to filter each of your json log lines and extract certain fields:

$ awslogs get my_lambda_group --query=message

This will only display the message field for each of the json log lines.

If you use tools like localstack, fakes3 or other, consider to change boto3 endpoint using --aws-endpoint-url or AWS_REGION env variable.

The required permissions to run awslogs are contained within the CloudWatchLogsReadOnlyAccess AWS managed permissions. As of 2020-01-13, these are the permissions:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Action": [
                "logs:Describe*",
                "logs:Get*",
                "logs:List*",
                "logs:StartQuery",
                "logs:StopQuery",
                "logs:TestMetricFilter",
                "logs:FilterLogEvents"
            ],
            "Effect": "Allow",
            "Resource": "*"
        }
    ]
}

• Fork the repository on GitHub.
• Write a test which shows that the bug was fixed or that the feature works as expected.
  • Use tox command to run all the tests in all locally available python version.
• Send a pull request and bug the maintainer until it gets merged and published. :).

For more instructions see TESTING.rst.

• https://aws.amazon.com/cloudwatch/
• https://boto.readthedocs.io/en/latest/ref/logs.html
• https://docs.aws.amazon.com/AmazonCloudWatch/latest/DeveloperGuide/cloudwatch_limits.html

Although, the most straightforward thing to do might be use --aws-access-key-id and --aws-secret-access-key, this will eventually become a pain in the ass.

• If you only have one AWS account, my personal recommendation would be to configure aws-cli. awslogs will use those credentials if available. If you have multiple AWS profiles managed by aws-cli, just add --profile [PROFILE_NAME] at the end of every awslogs command to use those credentials, or set the AWS_PROFILE env variable.
• If you don't want to setup aws-cli, I would recommend you to use envdir in order to make AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY available to awslogs.