Skip to content

iSCSI SAN nodes monitoring

to start moving onto other things, I closed the iSCSI storage setup ticket (#40131 (closed)) because the checklist was already pretty long...

so i split off the tasks regarding monitoring work, because that will require yet again more development. the tasks are:

  • evaluate monitoring solutions, presumably based on SNMP, but @munsond hinted at email support directly on the SAN?
  • if using SNMP, enable SNMP on all SANs SNMP polling is not supported
  • if using SNMP, figure out how to hook this into our monitoring (the Prometheus SNMP exporter is packaged in Debian, and the check_snmp plugin is standard Nagios) not happening, SNMP polling unsupported
  • deploy the SMcli command-line tool on chi nodes with Puppet
  • configure one Nagios hosts per SAN enclosure
  • configure ICMP probe checks for management and iSCSI IP endpoints (via NRPE)
  • configure a health status check per enclosure (via NRPE)
  • write a pager playbook (e.g. lsscsi and other tools we used to debug iSCSI on the client side, but also what to do in case of fire on the SAN, scenarios like total loss of a SAN should be considered as well)

note that the CLI guide says this about alerts:

Setting Up SMTP And SNMP Alerts

The storage array can be set up to send automatic e-mail alert messages to specified email addresses when specific events occur. View the current alert configuration settings using the following command:

SMcli -d -i -s -w -v -S

By default, all alert configuration settings are None.

The following example shows how to set the mail server IP and the sender address configurations for SMTP alerts:

SMcli -m 123.45.67.892 -F MyStorageArrayEvent@MyCompany.com

or

SMcli -m MyCompany.com -F MyStorageArrayEvent@MyCompany.com

An example of a command to set the email alert destination and specify that only event information is to be sent is:

SMcli -a email:MyCompanySupport@MyCompany.com 123.45.67.89 -I eventOnly

The following example shows how to set the SNMP trap alert configuration. In this example, the trap destination is 123.45.67.891. The storage array is 123.45.67.892, and the community name is public.

SMcli -a trap:public, 123.45.67.891 123.45.67.892

Then there's also the "Selecting The Event Levels For Alert Notifications" and "Configuring Alert Notifications" sections, but those, oddly, do not show how to do this from the commandline and only provide a instructions to clickety through the GUI, which is always a bad sign for us.

Edited by Jérôme Charaoui