Ability to add multiple alarms to a check

[sleenen]

It seems I can only add 1 alarm to a check, I'd like the ability to add more alarms to the same check (for example send a mail, but also send an SMS)

Is this already possible? If not, this would be a great feature imo.

[martinb3]

I think you might be describing notification plans, not alarms?

[sleenen]

I don't think so. I can add multiple alarms in the control panel, using a different notification method

[martinb3]

Do you also use different criteria for each alarm?

[martinb3]

You can send an email and an SMS using a notification plan on a single alarm, is why I ask. You don't need two alarms to send notifications to email and SMS.

[sleenen]

I do yes. For instance I'd like an email when filesystem is 85% full and 90%, but a text message on 90% and 95%.

[martinb3]

Gotcha. Thank you for clarifying that. I'll mark this issue as an enhancement :)

[sleenen]

Thank you, would be a great feature imo. More control over alarms that way.

[james-ai]

Is this feature on the roadmap? I'd be happy to put some time in if required.

[martinb3]

I'd be glad to review this and merge it, if someone wanted to work on a PR for it. Thanks!

[james-ai]

Update: Just realised some arguments weren't being passed through to the template which is why the below didn't work. Ignore this.

Where's the documentation on how the check/alarm yaml file in /etc/rackspace-monitoring-agent.conf.d/ should look?

I've tried the following to no avail:

disabled: false
label: Filesystem
period: 90
timeout: 30
type: agent.filesystem
details:
  target: /
alarms:
  alarm-disk-use:
    label: Filesystem use on /
    notification_plan_id: npTechnicalContactsEmail
    criteria: |
      if (percentage(metric['used'], metric['total']) >  ) {
        return new AlarmStatus(CRITICAL, 'Disk usage is above %, #{used} out of #{total}');
      }
      if (percentage(metric['used'], metric['total']) >  ) {
        return new AlarmStatus(WARNING, 'Disk usage is above %, #{used} out of #{total}');
      }
      return new AlarmStatus(OK, 'Disk usage is below your warning threshold of %, #{used} out of #{total}');
  alarm-inode-use:
    label: Inode use on /
    notification_plan_id: npTechnicalContactsEmail
    criteria: |
      if (percentage(metric['free_files'], metric['files']) < 20) {
        return new AlarmStatus(CRITICAL, 'Less than 20% free inodes');
      }

      if (percentage(metric['free_files'], metric['files']) < 30) {
        return new AlarmStatus(WARNING, 'Less than 30% free inodes');
      }

      if (percentage(metric['free_files'], metric['files']) > 30) {
        return new AlarmStatus(OK);
      }

[martinb3]

Here's some documentation on the YAML files:
https://developer.rackspace.com/docs/rackspace-monitoring/v1/developer-guide/#configure-agent-with-yaml

[james-ai]

Do we want to pre-define a number of alarms for users to switch on/off individually or do we want them to supply their own alarm templates which are then incorporated into the agent's yaml file?

[martinb3]

I think we should add some additional alarms via templates, but not provide control over switching them off and on. That way, once someone is ready to customize, they can simply just go with their own template and have access to everything (vs. tons of settings to wade through). How does that sound?