Hi,
We are busy with a project where we want to monitor latency for our devices under the Ping Class group using the default ping data source provided in Zenoss V4.
The problem is that most of our devices are using a 3G connection, where the first ping (icmp packet) from Zenoss only opens the connection to the device and gives a very inaccurate/bad latency (>2000ms). We can see three pings to the device from zenoss itself using tcpdump. We are assuming the first ping is the start of a new cycle from Zenhub, where Zenoss itself sends a ICMP packet to the device to see if it's down or UP (Which it will then create an event for). Then we assume the next two pings are coming from the Ping data source. The data source probably sends two pings to make sure there is a data entry made for the average rtt. The problem is, the two pings from the data source are 25-45 seconds apart from each other so we still get the high/inaccurate latency from where a connection is opened from the server to the device. We have tried looking into ways to decrease this amount, but we haven't been successful. What we are assuming (Yes, once again) the data source pings all the devices under the /Device/Ping group from the first device to the last one (We have about 1450 devices under the ping group) and then starts over for the second ping interval.
We have tried the following:
1. Tried to change the number of pings per cycle to 1,2,3,4,5 in the ping data source. Doesn't seem to have any affect in the amount if pings it sends.
2. Changing the parallel jobs from 10 to 20. Which doesn't help, I think these configuration settings don't apply to the ping data source?
3. Changed the pings to send in flight from 75 to 200. Same as above.
We can't find any documentation on the zenping data source.So any input/help or insight would be appreciated.
Server specifications:
CPU: Intel(R) Xeon(R) CPU E5640 @ 2.67GHz (4 Cores, 8 Threads)
RAM: 12GB
HDD: 2 X 250GB HDD's 10K in raid 1
Zenoss version: 4.2.0
Thanks 