From: hume@ug.cs.dal.ca
Has anyone tried using Cricket to monitor the amount of data going into
a cyclic buffer under INN?
I'm using cnfsstat -m (to get MRTG-style output) and grabbing the "current
fill" value. I'm basically just after the amount of data flowing into
each buffer during a five minute period.
I'm using the following rules:
<cricket-config>/INN/Cycbuffs/Defaults =
target --default--
short-desc = ""
long-desc = "Volume of data into cyclic buffer %short-desc%"
target-type = cycbuff
datasource buffusage
ds-source = "exec:2:/local/news/bin/cnfsstat -m %cb%"
desc = "Amount of data written to cyclic buffer"
bytes = 1
rrd-ds-type = COUNTER
targetType cycbuff
ds = buffusage
graph --default--
y-axis = "Bytes written"
units = bytes
draw-as = AREA
bytes = 1
While, for example:
<cricket-config>/INN/Cycbuffs/JUNK/Buffers =
target CYCB16
short-desc = "CYCB16"
cb = "CYCB16"
The problem is, I'm getting nonsensical answers when I look at the graphs.
I know the "JUNK" buffers receives about 100K of data per five minutes (at
the least), and when I run cnfsstat -m in a loop the numbers reflect that.
For example, from Cricket itself:
[07-Jun-1999 11:30:13 ] Retrieved data for cycb16 (): 293131264
[07-Jun-1999 11:35:12 ] Retrieved data for cycb16 (): 293263872
Those numbers are in bytes.
However, the grapher tells me during that period it received 256.20 bytes of
data. Which doesn't seem to be a number derived from the difference between
the above, or a division by eight, or anything else visible to me.
Anyone know what I might be doing wrong here?
------------------------------------------------------------------------
Having difficulty getting "in synch" with list members?
http://www.onelist.com
Try ONElist's Shared Calendar to organize events, meetings and more!
This archive was generated by hypermail 2b29 : Mon Mar 06 2000 - 19:00:52 PST