[collectd] Networking issues with collectd - RRDs show 'NaN' on server, but not on host.

[Hayden Katz]

Hi Guys,

I'm trying to setup collectd for data gathering on a distributed network
with 50+ hosts, and I've run into a hiccup on my third.

These hosts are all Centos 5 server systems, they're confirmed to have
different configured hostnames in collectd.conf, and I can see that the
RRDs are all dropping into individual folders.

The collectd server and the host I'm having issues with are on slightly
differing timezones (host is approx 1.5hours in the future.), but I have
half hour time differences on the other two hosts with no issues.

This is happening for ALL plugins on one particular server, but I'm going
to use the 'Load' plugin as an example.

So, without further ado, the RRDs in question are:

load.rrd on *server*

shortterm midterm longterm

1336456500: nan nan nan
1336457000: nan nan nan
1336457500: nan nan nan
1336458000: nan nan nan
1336458500: nan nan nan
1336459000: nan nan nan
1336459500: nan nan nan
.... etc

load.rrd on *client*

shortterm midterm longterm

1336522880: 6.6571428571e-02 5.1142857143e-02 9.1428571429e-03
1336522950: 1.8285714286e-02 3.6285714286e-02 0.0000000000e+00
1336523020: 4.9714285714e-02 3.7142857143e-02 1.4285714286e-03
1336523090: 9.8571428571e-02 4.7428571429e-02 6.5714285714e-03
1336523160: 5.2000000000e-02 4.4857142857e-02 2.0000000000e-03
1336523230: 6.1142857143e-02 4.7428571429e-02 8.5714285714e-04
1336523300: 6.4571428571e-02 5.1142857143e-02 5.7142857143e-04

Aside from the hostname, the configs are exactly the same between the 2
functional hosts and the non functional host.
I've used tcpdump to watch the data flow across the networks, and I can see
the update packets sent on the host and recieved on the server,

Any ideas why this is happening?

Cheers,

Hayden Katz.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.verplant.org/pipermail/collectd/attachments/20120509/77392e0e/attachment.html>