Add disk usage to docker input plugin

[R290]

Use Case

I'd like to be able to monitor Docker volume size over time using the Telegraf Docker input plugin.

Expected behavior

A configuration setting for setting the preferred objects to be monitored: containers / images / volumes / build cache (as defined in https://pkg.go.dev/github.com/docker/docker@v24.0.5+incompatible/api/types#DiskUsageObject)

Possibly a few additional settings to include or exclude objects based on their name, preferably with wild card support.

Actual behavior

None of the settings are available; disk usage is not implemented in the Docker input plugin.

Additional info

The Docker client of the existing plugin natively supports querying disk usage. For example, see:

• https://pkg.go.dev/github.com/docker/docker@v24.0.5+incompatible/api/types#DiskUsage
• https://github.com/moby/moby/blob/v24.0.5/client/disk_usage.go

This should make implementation trivial, although there needs to be agreement on the configuration scheme.

[powersj]

Hi @R290,

A configuration setting for setting the preferred objects to be monitored: containers / images / volumes / build cache

Are all of these actually interesting?

Adding a storage_objects = [] settings which could take the things we want to collect seems possible. Then if a user wanted to further exclude/include they can use metric filtering. I'd like to avoid needing to have an include/exclude for every object. Thoughts?

Thanks!

[R290]

storage_objects should in this case contain the type of object to monitor (containers / volumes / etc.)? Further filtering would be based on using tagpass / tagdrop on tags like container_name or volume_name or do we use something more generic like storage_object_name?

[powersj]

storage_objects should in this case contain the type of object to monitor (containers / volumes / etc.)?

yes that was my proposal

Further filtering would be based on using tagpass / tagdrop on tags like container_name or volume_name or do we use something more generic like storage_object_name?

I'm not sure till we see what data we get back, but either of those is what I was thinking about

Is this something you want to put up a PR for?

[R290]

Yes, I'll put something together and we can go from there.

[R290]

It seems that we are using quite an old Docker api version:

telegraf/plugins/inputs/docker/client.go

Line 14 in 67f9a0e

version = "1.24" // https://docs.docker.com/engine/api/

Somehow, the use of DiskUsageOptions is ignored in the older api version and all storage object types are always returned. Setting the the version to 1.42 (the latest at this moment) seems to resolve this. I don't know if we can just change this? All docker tests pass with 1.42.

[powersj]

Thanks for looking into this further.

I am good with updating the version. Looking through the Engine API version docs there are quite a number of changes, but it does look like they try to keep compatability.

[telegraf-tiger]

Hello! I am closing this issue due to inactivity. I hope you were able to resolve your problem, if not please try posting this question in our Community Slack or Community Forums or provide additional details in this issue and reqeust that it be re-opened. Thank you!

[telegraf-tiger]

Hello! I am closing this issue due to inactivity. I hope you were able to resolve your problem, if not please try posting this question in our Community Slack or Community Forums or provide additional details in this issue and reqeust that it be re-opened. Thank you!

[R290]

The PR is still open, so no need to close the issue.

[powersj]

Not sure why the bot tried closing this again :\ hopefully with us commenting on it, it will go away.