Long ping responses while doing heavy virtual disk I/O within a VM.

[Celso_Kern]

Hi,

I did this test in esx 3.0.2 and 3.5 and results were similar. Within a linux VM, i started:

dd if=/dev/zero of=/file bs=1024k count=1000

And observed a ping from an external/physical host to the VM, response times were up to 3200ms during this writes and eve packet loss was seen.

64 bytes from xx.127.39.xx: icmp_seq=123. time=260. ms

64 bytes from xx.127.39.xx: icmp_seq=124. time=3626. ms

64 bytes from xx.127.39.xx: icmp_seq=125. time=2611. ms

64 bytes from xx.127.39.xx: icmp_seq=126. time=1599. ms

64 bytes from xx.127.39.xx: icmp_seq=127. time=590. ms

...

172 packets transmitted, 171 packets received, 0% packet loss

Is there any customization to avoid it? Is it an expected behaviour in virtual machines?

Thanks

Celso

[Dave_Mishchenko]

Your post has been moved to the Performance forum

Dave Mishchenko

VMware Communities User Moderator

[jhanekom]

What type of underlying disk are you using? Direct-attached SCSI, FibreChannel SAN, iSCSI SAN or NFS?

[kjb007]

If you run the test again, and look at the esxtop output for CPU in guest, what do you see?

-KjB

[Celso_Kern]

Hi folks,

I am using direct attach scsi disks in a HP Proliant DL380.

esxtop says:

10:30:57am up 7 days 21:45, 66 worlds; CPU load average: 1.11, 1.11, 1.11

PCPU(%): 99.21 ; used total: 99.21

LCPU(%): 5.86, 93.34

CCPU(%): 0 us, 1 sy, 99 id, 1 wa ; cs/sec: 97

ID GID NAME NWLD %USED %RUN %SYS %WAIT %RDY %IDLE %OVRLP %CSTP %MLMTD

1 1 idle 2 0.89 47.99 0.00 0.00 100.00 0.00 0.01 0.00 0.00

2 2 system 6 0.00 0.01 0.00 600.00 0.00 0.00 0.00 0.00 0.00

6 6 helper 22 0.04 0.08 0.01 2200.00 0.14 0.00 0.01 0.00 0.00

7 7 drivers 17 0.00 0.01 0.00 1700.00 0.00 0.00 0.00 0.00 0.00

9 9 console 1 1.10 1.60 0.00 98.05 0.47 98.05 0.10 0.00 0.00

16 16 vmware-vmkauthd 1 0.00 0.00 0.00 100.00 0.00 0.00 0.00 0.00 0.00

20 20 Win2k3 6 92.57 96.21 0.01 495.84 8.65 0.00 0.39 0.00 0.00

22 22 Linux-pxeclient 5 0.85 1.46 0.11 493.27 5.86 0.00 0.13 0.00 0.00

23 23 Win2008 6 4.16 6.24 0.33 590.13 4.33 0.77 0.29 0.00 0.00

Best regards,

Celso

[kjb007]

You are definitely maxing out your CPU from the below post, which would be the reason your pings are delayed.

How many vCPU have you allocated to the VM and how many physical cores do you have per socket? The DL380 has 2 sockets, but are you using dual/quad core cpu's? If you're allocating more than 1 vCPU to the vm, I would reduce it to 1.

-KjB

[drummonds]

And have you verified that VMware Tools has been installed and the setup was run?

[Celso_Kern]

Installing vmtools has fix the problem!

Thank you,

Celso