Hi
I have a server which is being used as a control server to kickoff IO with vdbench to a number of VMs on a private network. The control server is multi-homed on both a corporate IP network so I can ssh to it (eth1 - 10.5.x.x), and the private network which the VMs are connected (eth0 - 192.168.0.x) to be able to connect to the VMs.
I am in the process of simply verifying the connectivity at this point with a very simple config file (vm.cfg below). It appears that when I kick off IO, vdbench is automatically choosing to use the corporate IP network (10.5.6.250) and the slave has no route or means to access this network.
./vdbench -v -f vm.cfg
09:27:40.078 Starting slave: ssh 192.168.0.10 -l root /opt/vdbench/vdbench SlaveJvm -m 10.5.6.250 -n 192.168.0.10-10-141222-09.27.39.832 -l lin-1-0 -p 5570
09:27:40.908 lin-1-0 : 09:27:35.575 common.failure():
09:27:40.909 lin-1-0 : java.net.SocketException: Network is unreachable
09:27:40.909 lin-1-0 : java.net.SocketException: Network is unreachable
Is there a means to define which IP to use in this instance? I've already tried swapping eth0 and eth1 naming but doesn't seem to matter. A case of the lowest IP address wins perhaps?
Many Thanks!
Chad
vm.cfg:
hd=lin-1,system=192.168.0.10,shell=ssh,user=root,vdbench=/opt/vdbench
sd=sdlin-1,host=lin-1,size=20g,lun=/dev/sdb,threads=8
wd=wd1,sd=(sdlin*),rdpct=50
rd=run1,wd=wd1,iorate=max,xfersize=(4k),elapsed=30,interval=1