Running Xymon server version 4.3.11 on Fedora Core 20 with rrdtool 1.4.8-8. A mess of HPUX/Linux clients running Xymon and Windows clients running MrBig.

Noticing that the rrd databases on only the Unix/Linux clients (including the Linux Xymon server itself) have holes in the disk, memory, process, CPU utilization, TCP/IP and network traffic graphs. The RRD databases have NaN values of varying time periods in the files - do not see consistent time holes. In some of the filesystem plots the holes do not overlap but are scattered around time-wise. Some of the plots/RRD files look okay such as CPU load and TCP/IP connection times. Not seeing anything in the xymon server logs indicating rrd issues.

I am drilling with google and so far the xymon threads on this seem to cover xymon_rrd type bugs that predate 4.3.11. I am also mystified why not seeing the NaN data splotches scattered around in the Windows RRD files. Would not expect this is network issue as the data splotches are showing up on the Xymon server's RRD files themselves - should be a short return hop sending any data out the Xymon server NIC back to itself. If there was congestion issues then would expect the NaN entries to occur across all received client transmissions to include the Windows clients.

Some of the threads mention using the no-cache tag with the rrddata and rrdstatus channels - have not done that yet. Threads were indicating this was a fix for plots not ~including the filesystem/disk plots. Will be trying this next - but does not explain why the Windows filesystem rrd's are populating properly.

Any thoughts other then the usual: upgrade, RTFM (umm, which M...), shoot up with massive amounts of heroin, discharge shotgun at head, switch everything over to Windows hosting. For any upgrade, what is the recommended next stable release? Seems I am always upgrading into a different set of issues...