Simulating events in EMC Smarts IP domains

As part of the process of testing an EMC Smarts (aka: IONIX) installation, it is sometimes desirable to cause the IP (aka AM or APM) component to report an event on demand that can be tracked up the stack into SAM.

The simplest choice, in my experience, is to force a "Down" and/or "Unresponsive" event.

By far the most obvious way to do this is to power a managed device down - but for practical reasons, this isn't always possible. An alternative I frequently use is to deliberately break the routing on the system hosting the IP domain such that access by the domain to the all the device's IP addresses is blocked.

A word of warning though: take care not to block an IP that is one of the next-hop addresses for the IP domain server itself, or any other address that will break your connection with it, or indeed it's connection with multiple devices. I'll assume you understand enough about IP networking in general, and your installation in particular to be able to check that you wont fall into these traps.

Also note: these instructions refer to domain servers running the RedHat Linux operating system. Similar results can probably be achieved on other platforms using the relevant tools.

The process goes like this..

  1. Choose a device to use in the test. One with a relatively small number of IPs is a good choice. Ideally: a Host rather than a Router.
  2. Make a list of the IPs discovered on the device using the command to invoke the getIPs method for the system, like this..
    • dmctl -s DOMAIN_NAME invoke CLASS::DEVICE_NAME getIPs
  3. Verify that all the listed IPs are safe to block access to (see the warning above)
  4. For each IP listed, check whether it is (a) managed and (b) currently responsive, like this..
    • dmctl -s DOMAIN_NAME get IP::IP-ADDRESS
  5. For each IP that is managed AND responsive, create a "broken" host route, like this..
    • route add -host ADDRESS gw 127.0.0.1
  6. After a while (depending on the polling interval configuration) the device should be reported "Unresponsive" and possibly "Down" (depending on the topology).
  7. Once you've checked the results of the test, cancel the bad routes by running the following command for each IP you ran the "route add" for, above..
    • route del -host ADDRESS gw 127.0.0.1
  8. After a while, the notifications should clear

 

Scroll to Top