v DEEP-2

What’s new

This is the first release of Monitoring framework through the DEEP-HDC project, providing new version of Openstack, Mesos and QCG probes:

Some of the main new common new features are:

  • Integration with the Orchestrator services and providers convention
  • Integration with DEEP IAM
  • Integration with DEEP CMDB

List of RfCs

Openstack probe

Bugs

  • [DPD-578] - No support for Glance API v2
  • [DPD-646] - Openstack Probe result log message
  • [DPD-647] - Openstack probe metrics sent to Zabbix
  • [DPD-648] - Openstack probe failure at IFCA site
  • [DPD-662] - Change service_type for nova
  • [DPD-677] - Openstack probe: missing bash script & log properties

Generic & Common Bugs

  • [Issue-83] - MesosZabbixProbe RPM does not work on CentOS 7.5
  • [Issue-81] - Document the use of the tag "providers.exceptions"
  • [Issue-10] - Connection to the Zabbix Wrapper fails
  • [Issue-6] - No error control when not finding the config file

Mesos probe

  • Updated to version 1.4
  • CMDB integration now is mandatory, so cmdb.location property now is mandatory in the configuration
  • Updated to provide new convention of groups and hosts (read below)
  • Groups and hosts semantics - This release changes the semantics of groups and hosts when sending metrics to the Zabbix server:
    • Since CMDB integration is now mandatory for both probes, provider name will be used as group name when sending metrics to the server
    • Host name will be the service ID
So for example, if we have two providers with the following services:
  • IFCA
    • "s1": Mesos
    • "s2": OpenStack1
    • "s3": OpenStack2
  • INFN
    • "s1": OpenStack
meaning that the provider IFCA has one Mesos cluster and two OpenStack instances running while the provider INFN has just one OpenStack installation.
In this scenario the OpenStack probe will create two groups (if they don't exist) named IFCA and INFN and it will register metrics under the host name s2 and s3 for group IFCA and s1for group INFN, meaning that metrics under the host s2 and group IFCA will be the ones related to the OpenStack1 instance running at IFCA provider. The same applies to the OpenStack2 instance in IFCA and the only OpenStack instance at INFN (in this case, the group would be INFN instead of IFCA).
The Mesos probe will do the same but given that there's only one Mesos cluster in IFCA it will create (if it doesn't exist) a group IFCA and a host with name s1 to send its metrics.