Losing connectivity to powerconnect 3048 after a while

Spanning Tree Protocol is a means of providing redundancy in a switched network, without creating loops that could lead to a broadcast storm.  By default, this is enabled on Dell switches (and, for that matter, most other managed switches on the market).  The reason you are likely having problems is that spanning tree is run against a port when it goes into an up state (such as when the cable is first plugged in, or when a system is rebooting/booted up for the first time).  This process usually takes about 45 seconds or more, depending on the size of the network, and no traffic (except for STP frames) will be forwarded by the switch.

Since you are using DHCP, your end systems are trying to request IP addresses from the DHCP server before Spanning Tree puts the port into forwarding mode.  Depending on how you've configured DHCP to renew leases, it often seems like the client systems cannot get a DHCP address.

If you've no plans to grow your network beyond this 3048, this issue may be easily dealt with by disabling Spanning Tree Protocol on the switch.  You can do this from the web interface by going to Spanning Tree > Bridge Settings.  There will be a pull-down menu to either enable or disable Spanning Tree globally on the switch.

If you are planning to grow the network in the near future, or you have redundant links between this 3048 and other switches, then you will probably want to leave Spanning Tree enabled.  However, you can bypass STP on a port-by-port basis using a feature called 'Fastlink'.

To enable Fastlink, go to Spanning Tree > Port Settings, and you will be given a list of ports on the switch.  There should be a column labelled 'Fastlink', which is enabled via a check-box.  Simply check the ports on which you want to enable this feature.  NOTE:  Do not enable Fastlink on uplink between switches and other internetworking devices (routers, hubs, switches, etc.) - Fastlink should only be enabled on ports connected to end systems.

We are going to order a 5212 gb switch but we are not creating any redundant network for now.

 

Do i need STP for this then? as it is not redundant.

That fastlink option is that a way to set the port in a forwarding mode? And i need to set it for all end-systems but the uplink to that 5212 switch is that right?

 

Ideally, if there is no possibility of a loop occurring in your topology, there is no need for the Spanning Tree protocol, as this will essentially add overhead to your traffic.   Spanning Tree is generally used to prevent loops, as well as provide redundancy. 

The 'FastLink' option allows the port to immediately go into a forwarding state.  FastLink bypasses the Spanning Tree algorithm (approx. 50 seconds) when the link initially comes up.  It is ideal to have this option enabled on any access ports (clients, servers, etc.) to prevent any issues connecting to resources when the system is initially coming up.  The Spanning Tree algorithm may hinder services such as DHCP or BootP since no traffic is forwarded until the protocol determines the topology. 

It would not be ideal to have this option enabled on any uplink to another networking device that supports the Spanning Tree protocol.   This will allow the protocol to run it's algorithm with neighboring switches and if there is a loop, perform it's functions of determining the best path, root device, etc.

Basically, enable the option on ports that only have 1 device attached to them.  In that case, there is no possibility of a loop, so there is no need to run the algorithm.

thanks

The problem is still there. i connected a 5212 thru one of the gb ports. and also i have a 3com switch connected.

My problem reoccurred even after disabling spanning tree. but i see spanning tree packets passing and using all the bandwith.

i have connected about 10 machines on it and when i add more the switch fails.

where do i have to look. because i do not know

g

As this issue is starting to become more and more one that will require some in-depth troubleshooting, I recommend you call into Dell Technical support for your region to get assistance.

One thing you might want to look at, regarding the sniffer, is where the spanning tree traffic is coming from.  It should list, in the traffic, the source - match that up to a MAC address.  If it's still coming from the 3048, then you *definitely* have an issue that should be dealt with through technical support.  If it's coming from another device, check that it has spanning tree disabled, as well.

I cannot honestly answer your problem about the 3048, but if you have the 5212, make sure the latest firmare is on it!  I had 2 5212's attached to a 3048(the primary switch) with the 5212 for server's, and the IT pc's.  Our network went down for 3 days and with many manhours wasted looking for the problem(we originally thought it a virus) we determined the 5212 was the culprit.  Within minutes of removing the 5212, our network was back up.  After hours testing by plugging the 5212 back in resulted in more network outtages.  A replacement 5212 switch kept the network up but caused everything else connected to it to get dropped off.  The switches were sent back.  Apparently Dell's tech support personnel are not aware of the issues with 5212's and 5224's as posted on this board, if so, I would not have wasted so much time on the phone with them, thinking there was some type of topology problem!  --  The initial setup of the switch worked fine for a month or so, after that, it was all down hill....