Locate write intense processes at ESXi

[JablesDE]

Hello everyone,

I've some performance problems on a Dell T610 running ESXi 6.0 U2 and I'm not able to identify reason.

This issue appears some weeks ago. I located a predicted fail to a harddisc and a defective RAID batterie.

Last week I had some maintanance windows and updated ESXi from 5.0, over 6.0 U1 to 6.0 U2.

Further I've updated firmware version for server and hardware devices.

Behavior:

The load on the server is mostly okay (20 - 30%), but at 9, 12 and 15 o'clock it's freaking out.

Storage, CPU, System, nearly all activities increases and clipping to 100 percent. VM latency raises so high

that database applications interrups and looses connection. After 7 - 10 minutes everything is over and the

system is running normal again.

Is ESXi running some processes every 3 hours, or does some write processes?

Thanks in advance!

Peter

[Kauy]

Hello, Peter.

Run esxtop on your servers to monitor performance at the time of a contention

Take a look at Chapter 7 of vSphere Monitoring and Performance - vSphere 6. There you will find some guidance on how to navigate on esxtop to find who which resource has been heavily used and by which VM/Process.

[sklaus]

Are you using NO CACHE raid controller like h200?

Write performance is crippled by cheap dell controllers and that can lead to a number of issues including yours

Please reply with controller info

Best regards

[JablesDE]

Hello Kauy,

I'll read read the documentation and protocoll the procedure.

Hello sklaus,

It's a H700, but write cache was deactivated automatically since the battery failed.

The battery was replaced friday. I've set write mode to "write back" and read mode to "adaptive pre read" now.

Thanks for that hint!

Many thanks to both for your hints! I'll report tomorrow if the issue is has gone, or deliver a new report.

Best regards,

Peter

[impranayk]

You can have a quick look at below article.

http://vmwareinsight.com/Tips/2016/4/5798826/vSphere-6-ESXTOP-Quick-Overview-for-Troubleshooting

[JablesDE]

Hello everyone,

it's sad but the reason for our disconnects/lags was the sheduled updater for Avira Antivirus Professional.

It updated our virtual servers every 3 hours. An simultaneously update on 13 VMs with additional load from our users was to much for the ESXi.

I configured the sheduler to update daily once at 4 a.m. and the problem was gone.

ESXTOP is a great tool to get a overviev about the running processes on an ESXi.

A thousand thanks to you all!

Best regards,

Peter