We occassionally have linux nodes that get fussy
and their cpu loads will steadily but quickly climb
to very high values. We believe this is related to
some automounter problems. Yesterday one node reached
a reported load of 993 for about 12 hours!
The problem is that this obscures the historical rrd
data in ganglia by expanding the range of the graphs
way too much and squishing all the real data down.
To fix this, I've done this on one grid:
1. shutdown gmetad
2. use "rrdtool dump file" on all the load rrds in
both the node subdirs and the __SummaryInfo__ dirs
3. manually zero-out all the bad data
4. recreate the rrds with "rrdtool restore file file.rrd"
5. restart gmetad
Questions:
A. Is there a simpler way to do this?
B. Is there an automated way to recreate the __SummaryInfo__
data after just changing the individual node data rrds?
Thanks!
=====
--
John Saalwaechter <[EMAIL PROTECTED]>
__________________________________
Do you Yahoo!?
All your favorites on one personal page Try My Yahoo!
http://my.yahoo.com
