Chore/collectd install prompt

[GlennChia]

Description of the issue

#274

Following the wizard, there is a prompt to monitor metrics from CollectD. If this is enabled, there will be errors downstream when configuring the agent with the created JSON template.

The error downstream is: Error running agent: Error parsing /opt/aws/amazon-cloudwatch-agent/etc/amazon-cloudwatch-agent.toml, open /usr/share/collectd/types.db: no such file or directory

This is error is also mentioned in an issue in the amazon-cloudwatch-user-guide.

The cause of this is because some systems don't have CollectD installed as explained in AWS documentation for retrieving custom metrics with collectd

Description of changes

Add a note in the wizard when users are prompted to enable CollectD from the wizard. This note informs users that they need to have CollectD installed (I tested this fix on Amazon Linux 2) or have the /usr/share/collectd/types.db file file created (I tested this fix on RHEL8). Credit to these fixes can be found in the issue in the amazon-cloudwatch-user-guide.

License

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

Tests

Amazon Linux 2

1. Install CollectD:

   sudo yum install -y https://dl.fedoraproject.org/pub/epel/epel-release-latest-7.noarch.rpm
   sudo yum install -y collectd
   

2. Assuming the configuration is stored in SSM parameters as mentioned in the associated issue, use the Systems Manager Run Command
   • Command document: AmazonCloudWatch-ManageAgent
   • Configure the location to the SSM parameter store with the JSON config from point 2
   • Run the command
3. This fixes the issue
   

RHEL8

1. I encountered some issues with CollectD install and went for the solution to create the /usr/share/collectd/types.db file file. Run the following:

   mkdir -p /usr/share/collectd/
   touch /usr/share/collectd/types.db

2. Assuming the configuration is stored in SSM parameters as mentioned in the associated issue, use the Systems Manager Run Command
   • Command document: AmazonCloudWatch-ManageAgent
   • Configure the location to the SSM parameter store with the JSON config from point 2
   • Run the command
3. This fixes the issue
   

[adam-mateen]

Since the CloudWatch Agent could fail to start, I'd recommend making the the "Note" into a flashy "WARNING" which is more likely to be seen by users:

yes := util.Yes("Do you want to monitor metrics from CollectD? WARNING: CollectD must be installed or the Agent will fail to start")

Another idea would be to modify the configuration validator to output a more helpful error message:

ERROR: collectd is configured, but is not installed. Please install collectd or change the config.

I'd like input from others on the preferred solution.

[jhnlsn]

It would be important not to break existing user workflows such as causing the config generator to fail if collectd is not installed properly. Someone could config the agent ahead of installing collectd.

What's the current agent behavior if collectd is configured but not installed?

[adam-mateen]

The Agent fails to start:
awsdocs/amazon-cloudwatch-user-guide#54

[SaxyPandaBear]

I agree with @adam-mateen on this that it should be made more obvious that there will be an issue if collectd is not installed prior to starting the CW agent.

1. This doesn't change any behavior - it just makes it clearer to the user that the agent will fail to start if collectd isn't installed
2. The one thing I'd be a little hesitant on is the length of the input prompt string now, but that's mostly a nitpick. For someone working on a smaller terminal, the formatting might get a little messed up with such a long string. That being said, if there's another prompt in the repo that is also very long, then I have no qualms with this.

[GlennChia]

Hi all, thanks for the review, I think the recommendation @adam-mateen made for option 1 fits the length (there's a longer prompt from func wantEC2TagDimensions) and is sufficient for the user to be aware of aware of the CollectD installation. Shall I proceed with the refactor for option 1?

[SaxyPandaBear]

I think the wording is sufficient. I'd say go for it