cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements
Join Customer Connection to register!
394
Views
0
Helpful
11
Replies
abarb002
Beginner

Interface Reset or Reconnect

Situation:

We don't have a back up generator/power source for our facility. So on weekends or major storms approaching, we have to shut down all servers (20) in our server room. Yes, we do this every weekend. 

 

Issue we are having is that when these servers are powered back up, there seems to be no connectivity to this device via the switch interface. Only after the cable is unplugged and plugged back in again does these servers regain their connection. 

 

We are using Cisco 9300 and 3850 switches on these servers. All servers are the latest Dell servers (540s etc). They are connected via trunk and access ports. I know trunks for more throughput but I need clarification on why the access ports as well. I think for network identification purposes (IP address).

 

Has anyone else encountered this issue?  

1 ACCEPTED SOLUTION

Accepted Solutions
Georg Pauwen
VIP Expert

Hello,

 

could be related to the Power Management options of your NIC. 

 

In the Power Management tab of your NIC, and check if the box next to 'Allow the computer to turn off this device to save power' is enable, if it is, uncheck that box.

 

Also, make sure all Wake On Lan settings are disabled (see the link below):

 

https://www.dell.com/support/kbdoc/nl-nl/000129137/wake-on-lan-wol-troubleshooting-best-practices?lang=en

 

View solution in original post

11 REPLIES 11
balaji.bandi
VIP Expert

not that we aware of the issue or encountered it.

 

what is the version of code running on both switches?

 

what status of the switch port - before you do shut and unshut or re-connect the cables.

is this an access port or trunk? all Servers same issue? what NIC cards these 

 

 



BB


*** Rate All Helpful Responses ***

They are trunks and access ports. The interfaces were in an up/up status with no issues. The logs do show that the interfaces are going up and down as the servers reset themselves. I'm running the latest IOS upgrades from Cisco. Not sure about the specifics about the NICs, only that they are installed on Dell 840s. It doesn't happen all the time, its just sometimes its almost as if the interface is in a sleep mode or something until the cable is unplugged and plugged back in again to regain connectivity. 

Do you have any other device in the switch having the same issue, apart from Dell Servers?

 

i suspect this may be some settings on the Server-side?

 

can you also post one of the port configuraiton from the switch to understand the issue?

 



BB


*** Rate All Helpful Responses ***

marce1000
VIP Advisor

 

 - Are you sure the switches are not impacted by the storm neither ? Anyway configure a syslog-server on all switches and use

a central syslog server to collect messages from the switches at all times. Look for anomalies during those problematic situations , if any.

 M.

I looked into the possibility of a storm, and I think this may be a switching loop issue. We have multiple connections to these servers for redundancy. So this may be the problem, so I will have to monitor this more closely. Or look for ways to prevent these loops. 

Georg Pauwen
VIP Expert

Hello,

 

could be related to the Power Management options of your NIC. 

 

In the Power Management tab of your NIC, and check if the box next to 'Allow the computer to turn off this device to save power' is enable, if it is, uncheck that box.

 

Also, make sure all Wake On Lan settings are disabled (see the link below):

 

https://www.dell.com/support/kbdoc/nl-nl/000129137/wake-on-lan-wol-troubleshooting-best-practices?lang=en

 

View solution in original post

Thanks sir for the info, I have printed the article and passed on to our server guys. 

 

Question, if there are multiple connections to the server, we have trunk and access for redundancy and more throughput, can this cause issues? Could this create a switching loop is what I'm asking. We've had to shut down our syslog server due to security concerns out of our control, so this a tool that has been taken away temporarily. 

Hi,

If a server has redundant links connected to the same switch, configure both ports as trunk ports and not one trunk and one access. Also, in order to avoid any loop, you want to configure the servers with high availability so, when one link fails the other one takes over. This works well in a Dell VM environment with no Portchannel needed. The other option would be to configure the server's 2 links and put them in LAG group and do the same thing on the switch (Portchannel). You can use LACP or mode "on"

I know the first option works really well with Nexus because interoperability between vendors can be a problem but since you using Catalyst, you may have to test this solution with a server or 2 to make sure it is optimal for your environment.

HTH

 

We don't have a back up generator/power source for our facility. So on weekends or major storms approaching, we have to shut down all servers (20) in our server room. Yes, we do this every weekend. 

Doing this every week is really painful. Can you put the servers on UPSs? Depending on the UPS size, you can connect one, 2, or 3 serves to one. These UPSs work well if these power outages are short (2, 3, or maybe even up to 10  min).

APC is a good option:

https://www.apc.com/shop/us/en/categories/power/uninterruptible-power-supply-ups-/network-and-server/N-15e4jmd

HTH

They are on ups, but the ups need replacing. We are running VDI on the VX-Rail systems, and according to Dell the system is very heat sensitive. So any prolong loss of power may cause major damage to the system if not destroy it all together. We are not a 24 hour facility so we shut down as a precaution over the weekend. 

Sir your solution was spot on. Had to change the power settings on the NICs on the servers. 

 

Thanks