Ping Sensor

Ping sensor fails with java.net.SocketException: No buffer space available

This error can occur when running discovery against large scope sets running on Linux where the IP interfaces are misconfigured.

Here is an example of the error that is seen in the ping sensor log.

2015-11-12 17:20:31,899 DiscoverManager [DiscoverWorker-23] 2015111217090862#PingSensor-10.8.21.23~10.202.123.36 INFO session.Ping - Ping failed for IP address 10.10.100.23 on port 22, error=java.net.SocketException: No buffer space available

This error can occur due to a misconfigured IP interface. Use the ip addr command to see how the interfaces are configured. Below is an example of a misconfigured interface.

bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER\_UP> mtu 1500 qdisc noqueue state UP
    link/ether 38:ea:a7:35:76:30 brd ff:ff:ff:ff:ff:ff
    inet 10.6.208.47/8 brd 10.255.255.255 scope global bond0

The /8 subnet is wrong. What is probably happening is ping buffers are filling up because so many ping requests are going out and the responses are not coming back.

Check the /etc/sysconfig/network-scripts/ifcfg-bond0 file to ensure that the values are correct.

[root@taddm network-scripts]# cat ifcfg-bond0  
DEVICE=bond0  
IPADDR=10.6.208.47
NETMASK=255.255.254.0
ONBOOT=yes
BOOTPROTO=none
USERCTL=no
NM_CONTROLLED=no
BONDING_OPTS="mode=1 miimon=100"
TYPE=Ethernet
PREFIX=8
DEFROUTE=yes
IPV4_FAILURE_FATAL=yes
IPV6INIT=no
NAME="System bond0"
UUID=ad33d8b0-1f7b-cab9-9447-ba07f855b143

The netmask can be set using NETMASK or PREFIX. Notice in the file above that since PREFIX is set last in the file, that is the value used even though NETMASK is set properly. Comment out or remove PREFIX (or change to 23) and reboot the server to resolve.