After installing new Switches | Intermittent lose of network connectivity

[treysha]

Hi, I have a very interesting problem that I believe expert network-ers will find chicken feed.

I have a network of about 100 users in my office. They are on a 10-yr-old 10/100Base-T network using 5 x 3Com Switch 1000 (24-port) unmanaged switches.

Recently, due to staff expansion, we extended our office to another level in the same building. I installed (using auto-configure) two new 3Com Superstack Switch 4228G managed switches on this new level, and on the old level replaced one Switch 1000 with a Switch 4228G. To put it simply, the deployment is now;

Old Level - 100 users - 4 x 3Com Switch 1000 (24-port) unmanaged switches & 1 x 3Com Superstack Switch 4228G managed switch (I would like to add that the 4 x Switch 1000s hold the clients while the servers are on the new Switch 4228G)
New Level - 40 users - 2 x 3Com Superstack Switch 4228G managed switches (all clients)

The two levels are linked by Fibre Optic cable using GBIC on the Switch 4228G.

The server remained at the old level, and there is no network connectivity problem on the new level, which is using two new switches.

However, on the old level, I notice that the 4 x Switch 1000s will intermittently lose network connectivity. ALL the clients on a particular Switch 1000 will lose connectivity, then ALL regain it back again. At times even two or three Switch 1000s lose connectivity at once. Just as I try to troubleshoot, they all regain their connectivity once again.

I wonder if it is the compatibility issues (strange, as all of them are 3Com), or is there something I do not know about using managed and unmanaged switches on the same network.

Can you help advise?

[Variable]

Maybe a loop. Probably the way they are set up. A loop occurs and the new switch sense it and shut down one portion of the loop.


http://www.oreillynet.com/pub/a/network/2001/03/30/net_2nd_lang.html

[juniper]

How are the switches connected? for example the 4228 that the servers are on should be your core switch and all other switches should plug into it directly (basicly only one hop to the servers) as well the core 4228 you want to force it to be the STP root bridge by changing its priority (think of STP as a layer 2 routing protocol, if a different switch becomes the root bridge then the STP hop would be 2 on some switches this could cause issues, I dont know a non-technical way to describe this) . also if these are superstack II 1000 series they are managed make sure you check there settings as a speed and duplex mismatch could be the issue (if someone set them static and the new switch is in auto it will be very flakey). also check for CRC and framing errors on the uplink ports, as well if the 1000's are managed check and see if the are reloading . Make sure you are using the uplink port on the 1000's as only one port is 100Mbps the others are 10 Mbps only

heres your user manual for the switch 1000 series



http://support.3com.com/infodeli/tools/switches/s_stack2/pdf/16900au5.pdf

another good thing to do is setup SNMP traps and install kiwi syslog to monitor and log the switch errors.