On 12/2/2016 8:35 AM, Alessandro Tinivelli wrote:

Hi everyone,

i’m trying to introduce Xymon in my new company, so i’ve installed xymon on centos7 with precompiled packets from epel repo.

All was going fine until a reboot: it happens that graphs are not updated: I see the correct “updated at…” but the line ends. More clear here

https://postimg.org/image/6a5viq3yn/

Restarting the service or rebooting the machine solves temporary the problem, but it comes back. The services seem to be up and running.

Tried disabling selinux, without results.

I did not find anything clear in the logfiles.

Version is 4.3.27-1.el7

**

This seems to be graphing-issue-Friday :)

This is usually an indication that there's data being lost somewhere in the pipeline, or there's a bug in xymond_rrd's caching mechanism.

Are you seeing any evidence of dropped messages either in your clients or in xymond_client's logs? Does xymond show any truncated/("bogus") messages in its status output?

If not, can you enable debugging on xymond_rrd (either -USR2 or just add --debug to the CMD line in tasks.cfg)? It should flush its cache within a second or two of showgraphs.sh triggering it in via the rrdctl.<pid> socket. If not, or if it's not happening in order, the RRD files can end up like that. (I've also seen this when there are multiple xymond_rrd processes running for the same channel (there should be two - one for status, one for data), which will end up writing over each other and clobbering RRA values. But that shouldn't normally be happening.)

Finally, can you run xymond_rrd in --nocache mode? If the graphs are subsequently clean, then that pretty much rules out message transmission issues elsewhere within xymon and we can focus on the RRD aspect.

Regards, -jc