IOwait on VMware Instances

[merz]

Hi @all,

I am running VMware ESX3.02 with lots of debian instances and NetApp NFS storage at a customers side. Now there are some instances experiencing

up to kind of 80-90% IOwait. There seems to be no clear reason for that. CPU overall looks fine, lightly used actually. Latenciesy measured on storage

also seem to be quite ok.

Now my question: Is there a VMware-internal possibility to get a better overview of what the instances CPU is doing or even waiting for ? Some kind of

per-process CPU utilization including per-process iowait statistics ?

Thanks in advance ... Frank ...

[dmaster]

Did you try the commands : esxtop and top from the service console ?

[drummonds]

Check the esxtop storage stats DAVG and KAVG first. There's information on those and many other counters on the page.

[merz]

Hi,

now I have results of esxtop and nmon for the client. nmon doesn't show this high IOwait and does not show any bottleneck for me. For interpreting

esxtop results I am not skilled enough, perhaps anyone could help:

ID	GID	NAME	NMEM	%USED	%SYS	%OVRLP	%RUN	%WAIT	%BWAIT	%TWAIT	%CRUN	%CSTP
1	1	idle	4	272.02	0.00	0.00	0.32	0.00	0.00	0.00	0.00	0.00
2	2	system	6	0.05	0.00	0.00	0.05	600.00	0.00	600.00	0.00	0.00
6	6	console	1	2.09	0.03	0.03	2.08	92.20	6.53	98.73	0.00	0.00
7	7	helper	13	0.00	0.00	0.00	0.00	1300.00	0.00	1300.00	0.00	0.00
8	8	drivers	9	0.01	0.00	0.00	0.01	900.00	0.00	900.00	0.00	0.00
40	40	affected-machine	5	3.19	0.17	0.11	3.04	485.72	15.74	500.00	0.00	0.00

Thanks for ya help ...

[dmaster]

I found this forum post maybe it's usefull..

VMware recommend that under "normal" operating conditions the CPU Ready value should not be above 5%

As per page 5 of this doc:

[drummonds]

Try reading the documentation I mentioned in the previous mail. If you find it unclear in some areas, I'm glad to update it. Obviously the storage section will be of particular interest to you.