Basicly, it's icmp only. We, like many other, utilize the tcp checking in
net_connect to do tcp checks against i.e. ssh and rdp in addition to icmp
checking the hosts.
Since we still wan't to do the tcp checks, we'll still need net_connect probe
with the same amount of profiles, and pretty much the same configuration.
Based on some testing I did earlier (but I might have gotten it wrong),
net_connect doesn't fail properly on the tcp checks if it's unable to check
them. However, I need to do more testing regarding this.
Furthermore, the new icmp probe has next to no config settings. What you can
configure is this:
- loglevel (default 1)
- timeout in seconds (defailt 2)
- number of packets (default 3)
- buffer size (default 64)
- interval in seconds (default 600)
- active (checkbox)
- interval in seconds
- buffer size
- TOS Format (default Decimal)
- TOS Type (default TOS)
- TOS Value (default 0)
So, they create a brand new probe that is gonna be the new ping flagship, and
it's utter garbarge when it comes to configuration.
What happened to configuration settings like..
- loglevel per profile? Debugging these probes with support for only global loglevel is a pain.
- No options to override fields like source, target and origin.
- No sample/alertafter settings to only trigger alarms if profile fails N intervals in a row.
- No support for retries within the interval.
- No support for custom messages on profiles.
- No obvious settings at least for monitoring packet loss.
- No way of selecting what QoS you want and don't want.
I mean, the new probe doesn't even support the same set of features related to
icmp that net_connect does.
And the memory usage. According to the doc, it needs: "Memory: 6 GB of RAM for
20000 resources, 8 GB for 30000 resources, 12 GB for 50000 resources." What on
earth are they doing? Storing the entire internet in the probe?
What is going on here?