2960 Switch and Unmanaged switch issues

[JaredRJ]

Hello all,

I work for an offshore drilling company. We currently have a platform that is having issues with an unmanaged switch attached to a 2960. Whenever they lose power, which is often, the devices connected to the unmanaged switch lose their IP address and connection to the gateway. 

Our networking guy says that it's an issue with the unmanaged switch, but I think it may be a configuration issue. Attached is the running config for the 2960. Any help would be much appreciated.

[marce1000]

 - Your post is not straight clear , if there is a power loss does that then not also include those devices and or how do they acquire ip addresses , static or dhcp ?

 M.

[JaredRJ]

The devices set to static seem to have no issue reestablishing the
connection, everything attached to the unmanaged is DHCP I believe. I'll
get more information.

[Flavio Miranda]

If you tell me how you guys fix the problme when it happens I will tell where is the problem!  Can you tell me? Imagine the site powered off and back again e then this device lost IP address configuration. What you guys do? Do you go to device and add the IP again?

[David Ruess]

Hello,

From your post it seems like your connection runs like this:

NETWORK <->UNMANAGED SWITCH <-> 2960

If that's the case and the 2960 has to go through the unmanaged switch to get out the rest of the network then that's correct that all devices connected to the UNMANAGED switch will lose connection and their IP addresses if the UNMANAGED switch loses power. You would need to either bypass that switch or put the UNMANAGED switch on a UPS to keep power when the building loses it.

-David

[JaredRJ]

I believe the unmanaged switch is in front of the 2960, I am gathering more
information but to my understanding there are only a few devices connected
to the 2969, including the unmanaged switch.
When the power fails, we restart our HMI box which has a static IP and that
one is fine. However the devices connected to the unmanaged switch seem to
all lose their connection to the gateway.

[Richard Burts]

I have looked at the switch config in the original post and find it pretty straight forward. There are 2 data vlans (100 and 101). Some ports are also configured with a voice vlan (131). There are several ports configured as trunks and appear to carry all of the vlans. There is a single SVI with an IP address for a management interface. Clearly this is a layer 2 switch with routing done somewhere else. I do not see anything in the config that looks like a problem.

The config suggests that there is some upstream device, almost certainly connected on one of the trunk ports, which provides routing and perhaps other services. And suggests that the unmanaged switch is down stream on one of the access ports. It would be helpful to get verification that this understanding is correct.

It would be very helpful to know where the unmanaged switch is connected. It would also be helpful to get an understanding of how the devices connected to the unmanaged switch get their IP addresses. And especially helpful to know what the recovery process is. If the devices lose their IP address it suggests that they are not manually configured. Are they using DHCP to get addresses? Is that upstream device the DHCP server? 

When those devices lose their IP address if you do a release and renew do they get an IP address? If you power cycle the device does it come up and get an IP address?

A better understanding of these things would be very helpful. 

I am going to make a guess about the problem. I believe that it is at least partly a timing issue. I am guessing that when power is restored the devices attempt to learn an IP address using DHCP, but do not get a response because the network is not yet functional. I am guessing that if you do release and renew that the devices are then ok. I am guessing that the unmanaged switch may complete its boot process quicker than the 2960 and there is not much that you can do about that. One thing that you could do would be to make sure that the port where the unmanaged switch connects to 2960 is configured with portfast.

[Leo Laohoo · Accepted solution]

Un-managed switch is "dumb".  If the power to both Cisco 2960 and the dumb switch bounce simultaneously, the dumb switch will boot up first and a few minutes later the 2960 will boot up last. 

I see this exact behaviour so many times with the first two "generation" of NICs.  Currently seeing this in clock controllers, printers, BMS.  The only way to fix this is to either bounce the ports leading to these clients OR rebooting them again.