intermittent switch management problem

[jkbarros]

I have several Cisco 2960 and 2950 switches across our campus that most of the time work perfectly.  However, infrequently, I lose communication with them, i.e. cannot ping the ip address assigned and therefore can no longer manage them.

Via console I cannot ping anything connected to or outside the switch nor can I ping the default gateway.  All traffic through the switch works including vtp and cdp traffic so it's not a loss of service, just switch management.  I can manage switches downstream from a switch that is not communicating.

A reboot sometimes brings it back but other times not.  All interfaces are enabled and not shut down. 

Uplinks are set statically to trunk mode.  All ip addresses are on the same vlan as the core switch / default gateway.

Any ideas?

[Gregory Snipes]

Sounds like you have a spanning tree problem. How large is your spanning tree diameter (the greatest number of hops between two layer 2 connected switches)?

[jkbarros]

I do not believe I have any more than 6 hops.  Is there a command, log or statistic that would reinforce this figure?

[glen.grant]

  Look at the arp table in the router and see if it sees the switch address .  Check to see if you have a duplicate address on the network that the switch uses.  Verify default gateway on switch is correct.

[jawad-mukhtar]

Check duplicate IP can cause that also

Jawad

[jkbarros]

From what I can tell there are no duplicates.  In the arp table on the core switch, the ip address of the 'down' switch does not exist.  I just re-drew our entire network map and to my knowledge the largest path has seven devices:

AH3B-AH3A-AH0A-Core-IH1A-IH2A-IH2B. 

Are there other spanning tree issues I should be considering? 

[haikalmesiniaga]

are u using fiber as your trunk ?have u check the trunk module ?