Add tags label to pve_guest_info metric

[ajgon]

Recent proxmox versions introduced tags for VMs. These provide useful for metrics as well, especially for configuring alerts i.e. something like pve_up{tags=~".*important.*"} == 0.

This PR adds tags to all metrics related to the VMs. Since .resource.get is used everywhere, this change should be backwards compatible.

[znerol]

Thanks for your contribution. This is a good start.

The tags label should only be added to the pve_guest_info metric though. The rest of the metrics should remain untouched.

[ajgon]

I think it makes sense for other as well - pve_up makes sense with example I provided before. For other metrics - there may be a use case, when we want to monitor CPU spikes only on few machines, but ignore on others (as they're duoing heavy duty tasks). So I think it won't hurt to add them there as well? pve_guest_info from what I understand, doesn't have much added value?

Edit: After hitting "comment" I think I noticed your point - basically using pve_guest_info as a reference for other tags? Something like:

(pve_guest_info{tags=~".*important.*"} * on(id) group_left(name) pve_up{}) == 0

[znerol]

Edit: After hitting "comment" I think I noticed your point - basically using pve_guest_info as a reference for other tags?

Exactly. See also this blog post for more info on joins.

Would you mind updating the example metrics in the README.md as well? How does it look like with multiple tags?

[ajgon]

Would you mind updating the example metrics in the README.md as well?

Done

How does it look like with multiple tags?

Proxmox returns them as string separated with semicolon (i.e. tags="tag1;tag2"). This comes directly as a string from API, and underlying library doesn't alter it in any way - so it translates nicely to label value.

[znerol]

Thank you @ajgon, released v3.4.0 which includes your valuable contribution.