Ping Timeout problem with Powerconnect 6248

Question

Dear All,

Please have a look at the topology;

Host X --> Switch A stacked with Switch B --> 10 GE Uplink --> Switch C --> Host Y

Lets consider Switch A and B are stacked and located as backbone and switch C is one of the edges. Host X can easily ping to Host Y or Switch C without any problem and also use any other services like DB, SMTP, DNS, etc. However, when Host X try to ping backbone Switch A, it is loosing ping packets. Ping responses irregularly.

In parallel if ping to switch C and switch A at different dos prompts, I can get continuously get ping responses from Switch C, but can not get continuously from switch A !..

I really wonder how this could be !... I did nothing in terms of packet precedence in the backbone. All the switches are 6248 with the latest image 2.0.0.12.

Does anyone had the same problem ? Is this normal or a kind of bug ?

I really appriciate if someone has an idea to solve it.

Best regards,

Sahin Akkargan

sentinel-master · Answer

Could you clarify the actual IP addresses your pinging, and the subnet masks associated with those addresses ?

Also clarify please if all switches (management addresses) and host devices are in the same VLAN, as this does sound like a IP/ARP or routing issue of sorts.

sakkargan · Answer

In switch A, there are several VLANs;

BACKBONE#show ip vlan

MAC Address used by Routing VLANs:   0019.B9AB.C272

           Logical
VLAN ID   Interface        IP Address       Subnet Mask
------- -------------- --------------- ---------------
2        vlan 2          10.102.10.9      255.255.255.248   ---> to firewall
6        vlan 6          192.168.6.254    255.255.255.0
16       vlan 16         192.168.16.254   255.255.255.0
20       vlan 20         192.168.20.254   255.255.255.0
21       vlan 21         192.168.21.254   255.255.255.0
29       vlan 29         192.168.29.254   255.255.255.0
100      vlan 100        192.168.0.254    255.255.255.0
111      vlan 111        10.20.30.1       255.255.255.0

and In switch C

BTSERVIS-1#show ip vlan

MAC Address used by Routing VLANs:   0019.B9A3.5946

           Logical
VLAN ID   Interface        IP Address       Subnet Mask
------- -------------- --------------- ---------------
20       vlan 20         192.168.20.251   255.255.255.0
29       vlan 29         192.168.29.251   255.255.255.0

Now, the problem is that any host behind switch A regardless of in which VLAN can easliy ping to any other host behind switch C, but get sometimes time out messages if ping to any VLAN gateways on switch A like x.x. x.254.

Also hosts connected to Switch A can easily ping each other, but they loose ping packets "depending on switch load" when they ping to associated VLAN gateways on switch A. Hope this clarify the situation, if you need further info like some commands output, I can send via e-mail.

Thanks again.
Sahin.

sentinel-master · Answer

Could you also clarify these points :

(1) Does the 10G link between the A/B stack and Switch C carry all VLANS - or just the two VLANS present on switch C (VLAN 20,29) - have you disallowed nay VLANS on the trunk......if it is a trunk ?

(2) Where is the default gateway (is it .251 or .254 in each VLAN) ,for the Hosts on switch A, and Switch ?

(3) Could you provide a "tracert" of a working cross-subnet ping & post here

(4) Could you provide a "tracert" of a non working/dropping ping connection & post here

Note : Just do Start -> Run ->cmd , hit return for a shell window from Host A, and then enter tracert IP.address

WWolkers · Answer

check the CPU load of the switch... These switches sometimes stop responding or the rtt goes up when CPU load goes up.

sakkargan · Answer

Thanks WWolkes,

You may be right, we loose ping packets only on working hours, during non-business hours ping responses seems OK.

As usual, during working hours cpu load is getting high instantly, however this does not prevent normal communication, we only loosing ping replies. In this respect, can we say that this is a built in property of the Dell switches ? It is amazing that I do not configure anything in terms of traffic shaping !..

I am not sure whether this is a kind of bug or very nice feature of Dell switches since I do not see any documentation regarding with the situation. What do you think about it? Have you ever experienced something like similiar ?

Also, I do not remember same behaviour with the other brand switches.

Thanks again, Sahin.

WWolkers · Answer

Things like ping times going up on higher CPU load happens on other equipment also, but mostly on the very cheap stuff. Usually related to bad firmware or underdimensioned hardware.

As long as it's only ping being dropped / ping latency going up I wouldn't be too worried. As soon as the telnet / ssh / webinterface stops responding you're getting in the higher cpu load area, that's when you have to get worried (been there already, at that point it's time to really check your config, and get in contact with Dell support for a firmware fix)

sentinel-master · Answer

Also try to make sure that the mangement IP address is in a seperate VLAN from normal user traffic/VLANs - as broadcasts, ARPS' etc in the management VLAN will cause the switch CPU to be forced to process the packet and waste time check to see -not for me.

By keeping the management VLAN away from normal users - you have a protected "quiet" VLAN just for management traffic & troubleshooting

Networking General

Ping Timeout problem with Powerconnect 6248

Was this post helpful?