cancel
Showing results for 
Search instead for 
Did you mean: 

Why is Stingray pinging my backend nodes when I've disabled the PING health monitor?

SOLVED
Highlighted
Contributor

Why is Stingray pinging my backend nodes when I've disabled the PING health monitor?

I've disabled the PING health monitor, but if I run tcpdump, I see that Stingray is still sending ICMP pings to my back-end nodes.  What's going on?

1 REPLY
Frequent Contributor

Re: Why is Stingray pinging my backend nodes when I've disabled the PING health monitor?

This is expected behavior; it's a consequence of the connectivity ( Feature Brief: Clustering and Fault Tolerance in Stingray Traffic Manager) performed by each traffic manager to verify that it is operating correctly:

Screen Shot 2013-05-15 at 10.32.13.png

Stingray will attempt to ping its front-end gateway to verify it has external connectivity, and it will attempt to ping some of the back-end nodes; as long as it can ping both the gateway and at least one active node, it will manage traffic normally.

If one of the two tests fails, Stingray will conclude that it has lost network connectivity and it will voluntarily drop any traffic IP addresses or traffic shares.  It will also broadcast a 'failed' status in the internal health status messages so that its peers in the cluster know that it has dropped its traffic ip addresses and shares.

You can change or disable the front-end 'gateway' test by configuring an alternative IP address to ping (see the global setting 'flipper!frontend_check_addrs').

You cannot disable the back-end node test, but if none of your nodes are able to respond to ping requests, then you can create a dummy pool that contains a node on localhost (127.0.0.1) - this local IP address should always respond to pings.

Note that if you do disable or work around these connectivity tests, then the Stingray device will be unable to detect front-end or back-end network failures and initiate a failover to a working machine.