Limitation

There is a limitation with the Ping Probe whereby the running of multiple instances of the probe on the same server is not supported.

If you try to do so, you may see the following types of messages written to the Ping Probe log file:

Debug: D-P_P-002-032: Packet not for our PID. Discarding.

The reason for this is that running multiple instances of the Ping Probe creates more than one instance sending Ping echos and expecting to receive Ping replies. (The multiple instances can be identified using the process ID or PID in the Ping packet.) As there is only a single channel to listen for incoming ICMP packets, this creates a race scenario between the two PIDs to process Ping replies. This may also make a particular probe instance unable to get actual remote machine status and may generate a false report (namely, reporting a host is down) when the other instance continues to discard Ping packets not for its PID.

Workaround

Alternatively as a workaround, you can install more than one network interface card (NIC) and bind each instance to a different IP address of the additional NIC using the Ping Probe's BindAddress property.

Note: For this workaround, the following conditions must be met, and the NIC in this context refers to either a physical NIC or a virtual NIC created by hypervisor:
  1. Each Ping Probe needs to bind to an IP Address of a different NIC. Binding to secondary IP addresses created on one NIC using a technique such as virtual IP or alias IP from O/S level is not supported.
  2. No other process or other instance of the Ping Probe are sending Ping packets using the same NIC binded by the running Ping Probe. There should be at least one extra NIC than the total number of Ping Probe instances to run, and this extra NIC should be configured as the default NIC for the other processes to perform Ping. For example, if you wish to start two instances of the Ping Probe, then three NICs are needed.