Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

I have a problem where my Layer 3 Dell PowerConnect 6248 Switch (main switch) will sometimes (that means sometimes it works, sometimes it won't) not get an arp response from a device behind Dell PowerConnect 2824 (edge switch).

The situation

- - - - -       ===============          ===============       - - - - - 
| client | ---- | Main Switch | -------- | Edge switch | ----- | Device |
- - - - -       ===============          ===============       - - - - - 
  1. If the Client is on same vlan as Device, there is no problem and traffic always flows.
  2. If the Client is on different vlan, then only sometimes traffic will flow. This switch routes between vlans (inter-vlan routing).
  3. Main switch broadcasts (many) arp packets, but seems like no response?!

    990 719.323028  Dell_xx:xx:xx   Broadcast   ARP Who has 10.11.0.43?  Tell 10.11.0.1
    

    because when looking at main switch's arp table, it doesn't know coressponding mac address:

    #show arp
    IP Address        MAC Address      Interface        Type        Age
    ---------------  -----------------  --------------  --------  -----------
    10.11.0.43       0000.0000.0000     vlan 11         Dynamic    0h  0m  2s
    
  4. Main switch knows to which port traffic has to be sent to reach Device. There is an entry in address table that shows on which port the device is, the only problem is the switch doesn't know that particular IP corresponds to this MAC address :(

    Address table:

    VLAN ID     MAC Address           Interface  
    VLAN 11     00:1c:c0:ef:e0:03       2/g5
    
  5. Edge switch is L2 switch and it appears that address tables are ok for this switch also.

  6. Some additional info about main switch:

    #show arp brief
    
    Age Time (seconds)............................. 1200
    Response Time (seconds)........................ 1
    Retries........................................ 4
    Cache Size..................................... 1024
    Dynamic Renew Mode ............................ Disable
    Total Entry Count Current / Peak .............. 68 / 99
    Static Entry Count Configured / Active / Max .. 6 / 6 / 64
    

Where is the problem?

Sometimes the traffic will just start flowing. Does the problem lie within the Main Switch or Edge Switch? Is it device configuration maybe? Maybe something with network cables? Main switch and edge switch is connected with fibre optics if it helps. Where should i look into?

Workaround

Creating a static arp entry solves the problem... currently I need only 5 arp entries for my devices, but this is just a workaround not a solution.

share|improve this question
    
Is the device just a server/pc or something else? –  Whisk Jan 19 '11 at 12:47
    
Could you paste in the output of a "show arp brief" command run on the 6200-series switch? –  Evan Anderson Jan 19 '11 at 14:38
    
client is PC, device is 1xPC, 4xNetwork Cameras. –  Janis Veinbergs Jan 19 '11 at 15:20
    
Having same problem here with M6348. When a client issues an ARP request to non-existent IP address, the client's address gets somehow internally removed from the switches routing table. This really looks like a switch bug.... –  ondra Apr 28 '11 at 11:08

2 Answers 2

up vote 2 down vote accepted

There is a known issue on the Dell firmware for the 6248. We experienced a similar issue that ARP entries were not being populated in the L3 ARP table. Sometimes they would work and other times they wouldn't.

If we did static ARP entries (MAC/IP) traffic would flow just fine.

Version 3.2.0.7 is what Dell indicated was the 'stable' version to back rev to. We are currently running 3.2.1.3 which has the symptoms mentioned above.

Hope that helps someone who may come across this!

share|improve this answer
    
oh, isn't that just freakin fun to find out... as i've been upgrading everything to 3.2.1.3... –  SpacemanSpiff Jul 13 '11 at 20:47
    
@SpacemanSpiff Uhm, but i'm running 3.2.0.7 and having these symptoms mentioned above. Could you please point out to any resource that approves that this is a known issue? –  Janis Veinbergs Jul 14 '11 at 6:38
    
ok, i feel better, 3.2.1.3 fixed alot of OTHER issues for me, i wasn't going to recommend to my last 10 customers to downgrade just yet –  SpacemanSpiff Jul 14 '11 at 14:18
1  
@Janis - Dell's support recommended 3.2.0.7 for our issue as it relates to ARP - probably a best guess. Another sign they indicated that it is the bug they are talking about is high cpu utilization in the forwarding process or H323 log entries (WARN) on the switch (but we didn't have the high CPU util). I did find it interesting that Dell quickly followed up with 3 more updates soon after 3.2.0.7. We're not going to back rev at this point, we're going to wait for the new firmware that will hopefully be out by 'end of July'. For now, the static entries will have to do as the workaround. –  Jeremy R Jul 14 '11 at 22:15
    
Thank you @Jeremy R. –  Janis Veinbergs Jul 15 '11 at 8:00

You should check the expected source of the ARPs -- the device itself.

Is the device configured with the IP address you listed? '10.11.0.43/16'? (is that subnet mask correct?)

Is the device receiving the ARP requests from the router? (tcpdump)

You mention changing VLANs. Are you changing them for the device or the client?

What do your routing tables look like? Is the main switch the default route for both the 'client' and the 'device'?

My initial guess is that your subnet masking on the device is /8 (255.0.0.0) when you need to set it to /16 (255.255.0.0) Either that or a misconfiguration of the IP might keep the device from responding to ARP requests.

Best of luck.

share|improve this answer
    
Thank you for comment. 1. I see that i actually get arp resonses (i had to turn admin mode on to see some more conversation when mirroring port). However that makes me think that i should search for the problem with my main switch configuration. 2. There is no problem with subnet - 5 devices act like this. 3. The VLAN is being changed for client. Simultaniously for clients on same vlan traffic flows, for clients other vlan - flows somtimes. –  Janis Veinbergs Jan 19 '11 at 13:55
    
By the way, when i create static arp entry, then it works.. :/ –  Janis Veinbergs Jan 19 '11 at 14:01
    
Anyway, i'll have to look at the arp's as close as possible to the devices itself, like you suggested. Routes are ok, everyone routes to main switch. (but different vlans, then the switch does inter-vlan routing stuff within iteself) –  Janis Veinbergs Jan 19 '11 at 15:23

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.