iostat cpu infos 🔗 link

	Hi,


	Hey ! You've got wait-io infos and don't use it :

[vmstat]
kthr      memory            page            disk          faults       cpu
r b w   swap  free  re  mf pi po fr de sr m1 m1 m1 m2   in   sy   cs  us sy id
0 0 0 1446640 537688 13 27 57  3  2  0  0  2  2  2  9  637  763  370  12  4 84
0 0 0 1446392 555136 115 75 605 5 5  0  0  2  2  2  8  489  775  155   6  3 91
[iostatcpu]
      cpu
us sy wt id
12  4  1 83
   6  3  2 90


	I'm terribly new to hobbit...could you tell me where I could modify  the way "CPU utilization" is made ?

	Another thing...I'd like for my user to have LA + CPU utilization  graph on "cpu" column....???

	grrrr. Could you please help me.


	I've found iostat cool informations returned from solaris/hobbit/ 
client :

[iostatcpu]
      cpu
us sy wt id
44 10  5 41
59 14 15 13


	I've found the waitIO from informations returned by aix/hobbit/client :
[vmstat]
kthr     memory             page              faults        cpu
----- ----------- ------------------------ ------------ -----------
r  b   avm   fre  re  pi  po  fr   sr  cy  in   sy  cs us sy id wa
1  3 324980   248   0   2   5 1615 3527   0 979 16601 1761 20 26 19 35
2  4 326939   319   0  14   6 653 1358   0 644 9467 4333 32 26  8 34

	
	I've found a cool definition in hobbitgraph.cfg :

[vmstat1]
         TITLE CPU Utilization
         YAXIS % Load
         -u 100
         -r
         DEF:cpu_idl=vmstat.rrd:cpu_idl:AVERAGE
         DEF:cpu_usr=vmstat.rrd:cpu_usr:AVERAGE
         DEF:cpu_sys=vmstat.rrd:cpu_sys:AVERAGE
         DEF:cpu_wait=vmstat.rrd:cpu_wait:AVERAGE
(...)

	Now, I'd like to have the IOwait column used to generate my CPU  
utilization graph....

On Wed, Jun 21, 2006 at 12:48:56PM +0200, Nicolas Dorfsman wrote:

[snip]

Currently not possible.

The I/O wait data in Hobbit was traditionally tracked by data provided
by the "vmstat" command. Since there is no I/O wait data reported by the
Solaris vmstat, there is no currently no place to store these data in 
the existing RRD (trend) databases - at least as they are used for data
from Solaris boxes.

What this shows is that the current setup which has a custom set of data
for each type of operating system was a poor design. It was inherited 
from the old Big Brother / LARRD setup - and it was a major mistake on
my part not to change it when I created Hobbit.

However, changing this now will require everyone to do a manual
export/import of data from their current vmstat RRD files into a new
file format. This is just not possible before the 4.2 release, so
I'm afraid you'll have to wait a while before you can track these data
by default.


Regards,
Henrik

Le 21 juin 06 à 13:07, Henrik Stoerner a écrit :

Currently not possible.

ok.

But.

I'm able to get these data from "iostatcpu".
iostatcpu is already returner by solaris client, and I've managed to  have the same from AIX :

nohup sh -c "iostat -t 300 2 1>$BBTMP/hobbit_iostatcpu.$MACHINEDOTS.$ $ 2>&1; mv $BBTMP/hobbit_iostatcpu.$MACHINEDOTS.$$ $BBTMP/ hobbit_iostatcpu.$MACHINEDOTS" </dev/null >/dev/null 2>&1 &
sleep 5
if test -f $BBTMP/hobbit_iostatcpu.$MACHINEDOTS; then echo  "[iostatcpu]"; cat $BBTMP/hobbit_iostatcpu.$MACHINEDOTS; rm -f $BBTMP/ hobbit_iostatcpu.$MACHINEDOTS; fi


How could I put theses figures into a RRD file, and then use the file  with [vmstat1] wonderful definition ?

Le 21 juin 06 à 13:29, Nicolas Dorfsman a écrit :

Hum.

1) I've modified hobbitclient-sunos.sh :

if test -f $BBTMP/hobbit_iostatcpu.$MACHINEDOTS; then echo  "[iostatcpu]"; tail -1 $BBTMP/hobbit_iostatcpu.$MACHINEDOTS; rm -f  $BBTMP/hobbit_iostatcpu.$MACHINEDOTS; fi   ### --> tail -1 instead of  cat

2) I've added (re-modified) hobbitclient-aix.sh :


nohup sh -c "iostat -t 300 2 1>$BBTMP/hobbit_iostatcpu.$MACHINEDOTS.$ $ 2>&1; mv $BBTMP/hobbit_iostatcpu.$MACHINEDOTS.$$ $BBTMP/ hobbit_iostatcpu.$MACHINEDOTS"</dev/null >/dev/null 2>&1 &
...snip...

if test -f $BBTMP/hobbit_iostatcpu.$MACHINEDOTS; then echo  "[iostatcpu]"; tail -1 $BBTMP/hobbit_iostatcpu.$MACHINEDOTS | awk  '{printf("%s %s %s %s\n"),$3,$4,$6,$5}'; rm -f $BBTMP/ hobbit_iostatcpu.$MACHINEDOTS; fi   ### ---> tail -1 and awk to print  the same infos in the same order

3) I've added iostatcpu in TEST2RRD :

TEST2RRD="cpu=la......iostatcpu=ncv"

and added in hobbitserver.cfg :

NCV_iostatcpu="cpu_usr:GAUGE,cpu_sys:GAUGE,cpu_wait:GAUGE,cpu_idl:GAUGE"

4) I've renamed [vmstat1] into [iostatcpu]


But...there's no rrd file created.  What else do I need to do ?

Le 21 juin 06 à 14:28, Nicolas Dorfsman a écrit :

[...snip...]
3) I've added iostatcpu in TEST2RRD :

TEST2RRD="cpu=la......iostatcpu=ncv"

and added in hobbitserver.cfg :

NCV_iostatcpu="cpu_usr:GAUGE,cpu_sys:GAUGE,cpu_wait:GAUGE,cpu_idl:GAUG E"

Just to be sure of my mistake :  informations sent by hobbitclient- xxx.sh are handled by by some special procs...so I can't use it to  add a new column or use an already generated info (like iostatcpu) as  I like ?????

	a quick 'google" answer.


	I've wrote a "cpuio" script which creat a new COLUMN with CPU  
utilisation including IO wait.

	The script is running on solaris an aix .

	I'm using NCV module to graph.

	I'll put it on deadcat some day.