0.8.7g : wrong ping method used for reachability checks...

[Canto]

Hello,

I am running cacti 0.8.7g and some weeks ago I moved my instance from a vmware virtualized machine to a physical server. I installed cacti from scratch on the new physical server and then dumped the sql tables on the running cacti to restore them on the new system (and transferring rrd files).
All looked good at first sight but yesterday one user reported that some of the devices they monitor had not been graphed since the move to the new server.

When troubleshooting, I realized that Cacti was detecting some devices as being down while they were in fact up (icmp and snmp correctly replying). When running tcpdump on the cacti server for some devices IP, it appears that Cacti is using UDP port 23 "ping" whereas it is correctly configured to use ICMP ping !

What could cause Cacti to use a check method different from the one that is configured ? I tried to change the settings and reapply the correct ones but still the UDP port 23 check ....

Any idea ?
Thanks

[gandalf]

Are you using spine? Which version?
R.

[TFC]

I have same issue.
Spine version 0.8.7g
And cacti 0.8.7g
I try to monitor my windows servers.
I can ping those servers by usin ICMP. (look at the figure 1 please)
But in devices section I see that Server is down (Look at figure 2 pls)
I dont know why I have this issue but when i look at cacti logs I see this:

08/24/2011 01:10:01 PM - SPINE: Poller[0] WARNING: Spine NOT running asroot. This is require if using ICMP. Please run "chmod +s;chown root:root spine" to resolve.

I run that command. But still I get this warning.
Why i see my servers as down?