2

After I rebooted one of my servers, Munin (which runs on another server) seems to have picked up some garbage response from it in the process of rebooting.

http://xcski.com/munin/xcski.com/linode.xcski.com-cpu.html

I would like to remove that garbage value, even if it means losing all the data for that node up to now. I tried removing the .rrd files in /var/lib/munin but all that did was remove the history from before the reboot. Is there a way to clean this up?

2 Answers 2

3

/var/lib/munin/hostname/* Should contain all the rdd files for that host.. If after deleting that you are still plotting garbage then maybe the munin-node on the offending machine needs to be restarted.

You can check to see if the node is sending garbage by telneting into the node service

#replace 4949 with whatever port you are using.
telnet bad-node-ip 4949

After you are connected enter

fetch thenameoftheplugin

You should see something like

system_cpu.value 123 
user_cpu.value 123 
etc

If you still see garbage then the node is still sending garbage so that would need to be inspected, not munin. (Start with the commands the script uses to output values to munin).

For additional information regarding debugging there is some info on the munin wiki

4
  • As I said in the question, I already tried deleting the rrd files. And the plugin is reporting normal results now. Dec 1, 2010 at 21:07
  • Have you changed the frequency munin updates the graphs?
    – Dan R
    Dec 1, 2010 at 21:19
  • why would I? Dec 4, 2010 at 21:56
  • @Paul, not saying you would. Just remember the docs warning against that because it would mess up RDD unless they were also configured to be filled at certain intervals.
    – Dan R
    Dec 10, 2010 at 20:59
3

Thanks to the munin-users mailing list, I have discovered the script "removespikes" which can remove transient values out of your rrd files. It worked beautifully.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .