Sleeping NIC issue across multiple versions of CentOS

[sszczesn]

Hello,

I'm having kind of a weird issue. I've got several ESX and ESXi servers

with multiple vms on each one, running various flavors of CentOS 5 and

various versions of DirectAdmin installed for web management. The issue

I'm having is the VM will loose connectivity to the internet unless I

console in and ping out from command line. As soon as I do that, the

connection is good to go for anywhere from 5 minutes to a couple of

hours, then the issue repeats. I have put a temporary fix in place by

setting up a cron job to ping google.com every 5 minutes to keep the

network connection alive.

Has anyone seen or experienced this? I can provide any additional

information, and I thank you in advance for any assistance you can

provide.

Thank you,

Scott

[sszczesn]

Any help at all is appreciated.

[golddiggie]

Never had this issue with any release of CentOS (4.x or 5.x) on ESX/ESXi hosts... Have you assigned fixed IP's to these systems or are they getting IP's via DHCP? What's the host hardware (include NIC's)? How are the vSwitches configured (especially for the VM traffic)? I've been running CentOS 5.x VM's on ESXi (first 3.5 and now up to 4.1) with fixed IP's without issue. I have run the same configuration in a production environment as well without issue...

You have a misconfiguration someplace since this is definately NOT normal activity...

VMware VCP4

Consider awarding points for "helpful" and/or "correct" answers.

[runguyenhuu]

Hi everyone !

I have same problem with CentOS 6.4,6.5 and 6.6. Have any solution to fix this problem ? Thanks you very much !

Hi sszczesn ! Do you have solution for this problem ? Can you tell me ?

[RobAtHomeNet]

I'm having a very similar issue...

Details on the environment:

• 1 vCenter server 5.5 u2b - build 2183111
• 2 balanced clusters with 2 hosts each
  • All hosts - Dell PowerEdge R715 - running ESXi 5.5 u1 - build 2302651
  • Cluster exhibiting the issue is showing CPU, RAM & iSCSI storage running ≤ 25% utilization with ≤ 24 guests running
• Guests:
  • Multiple guests running Windows 2008-R2, 2012 & 2012-R2 - hardware version 10
  • Multiple guests running Ubuntu 12.04 & 14.04 - hardware version 10
• Networking (not fully sure I have all the details here)
  • 1 virtual distributed switch, using 94/133 (71%) of the ports, running 9 networks across the 2 clusters

Description of my issue:

The issue only seems to affect my Ubuntu machines, as near as I can tell.  I only had one Ubuntu machine on the VLAN (212) that was showing the issue.  However, in order to test a theory, I set up another one (same specs & OS level) on another VLAN (214) which is our DHCP VLAN.  The guest never lost connectivity to the network while on VLAN-214.  I then moved it to VLAN-212, with the other Ubuntu server that was having the issue, and it lost the connection within 15 minutes - basically, the time it took me to grab a bite to eat and come back.  Since then, I've kept a CMD window open on my desktop and I ping the guest's IP about every 5-10 minutes.  I'm seeing the connection drop with that 5-10 minute window!  To get it back, I merely have to go into vSphere, edit the guest's NIC settings by "unplugging" the virtual network cable, save it, open it back up and reconnect the virtual network cable and save it again.  The fix takes about 30 seconds to employ but until then, the server is useless.  I've also been able to console into it, via vSphere, and ping out in order to get the NIC to wake up again.  I'v been able to prevent the server from dropping off the network by running a continuous ping (ping -t 192.168.1.200) to it from my desktop.

I'm curious if I have a misconfiguration somewhere.  However, if that's the case, why hasn't it affected the other VLANs?  (I think I'll move my test VM to another statically-assigned-IP VLAN and see if it happens there too.  Like I said, it had not happened on the DHCP-VLAN.  I'm no expert when it comes to VMware; all I've learned came from trial & error with some reading around the web.  So, any help with troubleshooting would be greatly appreciated!  I'm really trying to avoid setting up a cron job to ping out from the affected guests every minute or so...