Can’t push log data to CloudWatch Logs with awslogs agent? We can help you.

Here, at Bobcares, we assist our customers with several AWS queries as part of our AWS Support Services.

Today, let us see how we can troubleshoot this.

Can’t push log data to CloudWatch Logs with awslogs agent

In order to begin, we need to make sure that the awslogs agent can connect to the CloudWatch Logs API endpoint.

In addition, the configuration must have the following:

1. Internet connectivity
2. Valid security group configurations
3. Valid network ACLs

Fingerprint issues

While we configure data to push to CloudWatch we set a file’s path. We review the header lines of the source log file.

Suppose the first few lines are blank or contain non-event data that stays the same. This means, there is an issue with the log-identifying hash.

In this case, we update the file_fingerprint_lines option in the agent configuration file.

In addition, we need to ensure to specify what lines in each file we use to generate the identifying hash.

Check the awslogs log file for errors

Initially, we review the /var/log/awslogs.log log file. We might come across the errors below.

• Permissions errors:

NoCredentialsError: Unable to locate credentials

In case we didn’t add an AWS IAM role to the instance, we need to create and attach it.

However, if we did add an IAM role to the instance, we need to update the IAM user credentials in the /etc/awslogs/awscli.conf file.

ClientError: An error occurred (AccessDeniedException) when calling the PutLogEvents operation: User: arn:aws:iam::012345678910:<role/user>/<iam-user-name> is not authorized to perform: logs:PutLogEvents[…]

Here, we configure the IAM role or user with the necessary permissions for CloudWatch Logs.

• Timestamp errors:

Fall back to previous event time: {‘timestamp’: 1492395793000, ‘start_position’: 17280L, ‘end_position’: 17389L}, previousEventTime: 1492395793000, reason: timestamp could not be parsed from message.

We make sure that the log events begin with a timestamp.

Then we check if the datetime_format in /etc/awslogs/awslogs.conf matches the timestamp format of the log events.

If not, we change the datetime_format.

No file is found with given path ‘<PATH-TO-FILE>’

All we need to do here is to update the log file path in the agent configuration file to the correct path.

Caught exception: An error occurred (InvalidSequenceTokenException) when calling the PutLogEvents operation: The given sequenceToken is invalid[…] -or- Multiple agents might be sending log events to log stream[…]

It is not possible to push logs from multiple log files to a single log stream. So, we update the configuration to push each log to a log stream-log group combination.

Other awslogs issues

Sometimes, the logs may stop pushing after a log rotation. In that case, we check the supported log rotation methods.

If it pushes briefly after the restart, we check for duplicates in the [logstream] section of the agent configuration file.

Suppose the awslogs.log log file takes up too much disk space. Then we need to check the log file for errors and then correct them.

If it contains only informational messages, we specify a lower logging level.

Further troubleshooting

For further troubleshooting, we note the instance-id. Then, we collect and review the following based on the configuration.

Yum installations:

• yum version

  $ yum info awslogs
  $ yum info aws-cli-plugin-cloudwatch-logs

• /etc/awslogs/awslogs.conf file
• /etc/awslogs/awscli.conf file
• Other relevant files in /etc/awslogs/
• /var/log/awslogs.log file

Script-based installations:

• The awslogs version:

  $ /var/awslogs/bin/awslogs-version.sh

• /var/awslogs/etc/awslogs.conf file
• /var/awslogs/etc/awscli.conf file
• Other relevant files in /var/awslogs/etc/
• /var/log/awslogs.log
• /var/log/awslogs-agent-setup.log

For rotation-related issues, collect and review:

• A snippet of the source logs
• A list of the monitoring target directory’s contents:

  $ ls -la <Monitoring-Target-Directory-Path>

[Need help with the troubleshooting? We are here for you]

Conclusion

In short, we saw how our Support Techs troubleshoot the log data query for our customers.