ping some VMs with long response time or packet lost

[olano]

Hi,

I just met some strange things. Many VMs in our environment can't be connected with RDP. We have investigated and tried to ping these VMs. And we find the ping has long response time (time>3000ms) or some packets lost which shows some network problems. But the VMs aren't located on a special host. Even in one ESX host and in one vswitch, some VMs are ok, some not.

We found a workaround to migrate the impacted VMs in the cluster or reset the VM. Because these happens frenquently. Is there anyone know the possible cause?

Lan

[jguzmanr]

I can think of two causes......Are you using any kind of load balancing other than virtual port based? If you are, make sure it's the same on the vswitch and the pswitch. I once had a problem when one vm wouldn't respond, and it was because the port on the pswitch was on a wrong vlan. So all the vm's on the other pnic's were fine, but this one vm wasn't.

[olano]

Hi jrguzmanr ,

our load balancing is based on virtual port ID. Is it OK?

Lan

[danm66]

are you using vlan's? If so, what does your switchport config look like.

[olano]

Hi,

We have no VLAN on this vswitch.

vSwitch configuration

. pNICs : two, hard coded to 1000Mb/Full

. NIC Teaming:

.Load Balancing: Route based on the originating virtual port ID

.Network Failover Detector: Beacon Probing

.Notify Switches: Yes

. Rolling Failover: No

.Traffic Shaping:

.Status: Enabled

. Average Bandwidth: 819200kps

.Peak Bandwidth:1638400kps

.Burst Size: 1024000KB

.Security:

.Promiscuous Mode: Reject

.MAC Address Changes: Reject

.Forged Transmits:Reject

Lan

.

[ericsully]

There is a free Response Time Monitoring tool that could be of use.[http://www.real-user-monitoring.com]