Internet down every day - catalyst 2960

[fmylo]

Hi everyone,

I have problem with my setup which is ridiculously easy but still cannot fix it. So, the plan goes like this:

Isp router - - > 2960 switch (L2) - - > 9 access points

I have connected the ethernet line from isp router to fa0/1 and all the access points to fa0/2 - fa0/10. I made a separate vlan 99 for these ports (fa0/1-fa0/10) and configured them as:

switchport mode access

switchport access vlan 99

I gave the vlan 99 an ip 192.168.1.253 255.255.255.0 and my default gateway 192.168.1.1 (the isp router).

The issue: The internet connection is stable for only about 1 day and then goes down. And I have to restart the router to keep the connection alive. Next day the same and so on and so on...

I contacted with the isp and they didn't find any problem with my dsl line up to the router. So I suppose that I have messed it up with my configuration!!

Consider though, that when connection is OK I can ping from a single room both the default gateway and the other rooms (access points).

Can anyone help? Thank you in advance

[Richard Burts]

This is an odd symptom. To be sure that I understand correctly:

- it works and all connected devices are successful in Internet access

- after about one day it stops working

- you restart the router and it works again

So here are a few questions:

- you do not do anything to the switch. you only restart the router?

- when it is working can connected devices ping the default gateway address? (I would certainly hope so)

- when it is not working can connected devices ping the default gateway address?

- on a connected device, when it is working, can you post the content of the arp table of the device (arp -a or appropriate other command if not Windows)

- on a connected device, when it is not working, can you post the content of the arp table of the device (arp -a or appropriate other command if not Windows)

If it works ok and then does not work it suggests that something is timing out.

If the fix is to reboot the router and to not do anything on the switch or on the connected devices then it suggests that it is some issue on the router.

I can think of 2 types of things that might have time out issues that could relate to this:

- something that impacts basic IP connectivity. If arp is not working when the problem happens it is some type of basic connectivity issue. 

- something that impacts address translation. If arp still works when the problem is happening then it suggests that it is a nat issue on the router.

[fmylo]

Rick thank you for the reply! I will check the things you told me and I will post the relative content once I get on site.

For sure:

-Only router is restarted to accomplish connecticity. Nothing done on the switch

-When it is working connected devices ping the default gateway

Note this : Before my setting up the Cisco switch the connection worked normally with an unmanaged switch (during the whole past winter).

One of the things I was worried about is if there is any time out mechanism on the switch itself affecting the router as well. I am new to this so please forgive my ignorance if this sounds strange.

Anyway, I will post the content for you having a better view. Thanks again

[Reza Sharifi]

Note this : Before my setting up the Cisco switch the connection worked normally with an unmanaged switch (during the whole past winter).

So, the issue appears to be the new switch (2960) you have installed.

I have connected the ethernet line from isp router to fa0/1

When this happens, can you connect to the switch and see if port fa0/1 is down or up? If it is down at that time, can you do a "sh" and "no sh" to reset the interface without rebooting the router?

Also, can you post the output of "sh int fa0/1"?

Anything in logs of the switch?

HTH

[paul driver]

Hello
Even though you have mentioned this issue doesnt happen with an unmamaged switch Im wondering if the ISP dsl port is going to sleep, In the isp modem wan settings is there an option to make it "always on"

Also try putting all the switchports into stp portfast.

tnt x/x
spanning-tree portfast

[Richard Burts]

Thanks for the additional information. If it worked consistently on an unmanaged switch and has the problem when using the 2960 then it does suggest that it is something on the 2960. But I am puzzled what that might be. If all ports are assigned to the same vlan as access ports on a managed switch then it should act the same as an unmanaged switch (there is nothing about the Ethernet frame from a vlan port on a managed switch that is different from a frame from an unmanaged switch). Things only change if there is a trunk port or if there is more than one vlan used.

The suggestion to shut and no shut the switch port connecting to the router is an interesting suggestion. I would be very interested in the results of this. And the suggestion  to check the logs of the switch is certainly a good idea.

The suggestion for portfast is also interesting. If things work for most of a day before the problem occurs I am not optimistic that portfast will make a difference. But I think it would be in a category of "can not hurt and might help" so I would advocate for trying this.

I continue to wonder about the timing of the problem, if things work for most of a day and then stop working then what changed? Most of the things that I can think of on a 2960 that time out do so measured in minutes or perhaps a few hours. I can not think of what would time out after running for most of a day. Perhaps posting the running config might shed some light on this?

[paul driver]

---

@Richard Burts wrote:

. I can not think of what would time out after running for most of a day. Perhaps posting the running config might shed some light on this?

---

DHCP renewal is one possibility. by default lease is set to 24rs, so half lifetime is usually when a host tries to renew so it could be that dhcp client is trying to renew and it times out before its able to succeed, as such it expires as/when lease period has been reached.

[Richard Burts]

@paul driver DHCP is an interesting possibility. And the time frame for the problem seems to fit. So now I am wondering what there might be about changing from an unmanaged switch to a 2960 would impact DHCP. And I come up empty. Do you have suggestions about that?

[fmylo]

Thank you all for the replies,

In the next couple of hours I will try everything and post all info requested. I will also try to delete vlan 99 and leave all the ports on the default vlan 1. Security (at least now) is not an issue. 

[fmylo]

Hi again,

So... No ping while not working. Fa0/1 shut and no shut didn't work. And also the logs just showed all the interfaces go down and then up again.

What I did:

I put everything on vlan 1 with no interface ip address and deleted vlan 99.

I also changed the dhcp on the isp router from 2 hours that it was to 3 weeks and restarted it. Everything works fine until now but I will check again later and finally post my content (sorry I didn't have the time once I made the changes). I will be back with the new data in 2 or 3 hours hopefully

PS: I will also try portfast 

[Richard Burts]

Thanks for the update. When it was not working and ping did not work, did you check the arp table on the PC? It occurs to me that when it is not working it might be helpful to see the output of ipconfig (or similar command if not on a Windows PC) as well as the arp table.

[fmylo]

I was ready to post! This is the ping,arp -a  screenshot. Now it's working though, so I don't know if ipconfig will help you

[Richard Burts · Accepted solution]

Thank you for the new information. The arp output is quite helpful. And at this point ipconfig would not be helpful. I see 2 important things in the arp output:

1) there is no arp entry for 192.168.1.1. This is why the ping is failing with transmit failure. 

2) the arp entries show that the PC is currently using IP address of 169.254.254.145

This confirms that the suggestion about the problem being related to DHCP is correct. It appears that the PC has lost its valid IP and is using the 169.254 address because it is not able to renew its DHCP lease. So the question becomes why is the PC able to get an IP address initially but not able to renew the address? Perhaps seeing the running config might help us understand this?

[fmylo]

Thanks Rick! I also post the run conf but it's the current one with the changes I made this morning and for now is still working.

[fmylo]

Unfortunately, I will be away from the switch for the next hours but I will be online if I can share anything else. Consider though, that if in about 10 hours the connection is still up I will consider the issue solved. Many thanks for your super help!