SS 3300-Problems / TCP/IP and Appletalk

[SteelBurner]

Hi! I have two 3300s connected with a matrix cable. Periodically (once or twice a week) all equipment connected to the 2nd unit suddenly loose connectivity. Equipment connected to the 1st unit seem fine.

I have PCs running TCP/IP on 10/100, Apple MACs running TCP/IP and AppleTalk on 10/100 and some printers at 10/100 or just 10. NT server, W2K server and DSL router are connected to the first unit. NIC drivers are upto date and latest SPs on the Windows machines. I'm not familiar with MACs...

Both switches have the latest FW applied a few weeks ago but didn't help - it seems that problems are getting worse! Spanning tree protocol (STP) is disabled on both - I found that tip on the 3com and Apple website.

Most settings are system defaults, no v-lans, ports set to auto negotiate. I do see that forwarding is set to 'Intelligent' rather than store-and-forward, but I dont feel thats the issue - correct me if I'm wrong!

A manual power cycle corrects the problem and equipment can begin immediately communicating with the network again.

Any suggestinons?

Thank you!

SteelBurner

 

[AlphaZ]

Have you moved any of your MAC/Apple Volumes from an old server to a new server?
OR
Changed a NIC card in one of your servers?

The Mac's store the phyical layer path in the file resource fork. So if you move a volume to a new server and open a file there the Mac workstation will flood the network looking for the orginnal server the File Volume was on.

That was one!

Now do you have the ports on your switch set to auto or are you setting the speed for full duplex?

For the Mac's go to Apples website and get a copy of "apple ethernet speed & duplex tool (6.2.1)". This will help stop some of your break down. FYI most of the older and some of the newer Macs with built in NICs will only do half duplex, if you push them to full duplex they will go faster, but not well.

That was two!

If this helps, there are more tweaks that can be done.


BTW: It is always better to group the MAC's in the same switch. Also create a default vlan on you switches.

 

[SteelBurner]

AlphaZ,

thank you for your reply.

Indeed, I have moved volumes to a new partition on the server so this might be a problem. I'm using MacServer 8 IP on the NT-Server to make shares available to the Mac's. The old shares have been deleted and set up again on the new partition.

If this might be my problem, what can I do to stop the Mac's to look for the original path?

I also had to change the server's nic as it failed to work any longer, but at this time I already had the communication problems for about 3 weeks. It was a built-in nic on the Fujitsu server board. It is disabled now and replaced by a 3c980 server nic. Problems did not stopped after that.

All ports on both switches are set to auto. Switch diag says machines are connected with 100/FD (just so old printservers are running at 10/HD).

While I'm writing this the network seems to run without any problems. I removed the matrix cable and use a simple patch cord to connect the switches. It's not quite fast but works ;-)

I'll speak to the Mac guy to check for "apple ethernet speed & duplex tool (6.2.1)".

Also I'll group the Mac's on the same switch with the server.

Hope it helps, thank you anyway,

SteelBurner

 

[AlphaZ]

Another strange thing with Macs & 3Com switches. If you don't have all of the ports in a vlan that you create, you may find some interesting things. Like the Macs & PCs creating their own VLAN with each other and the servers.

As far as fixing of the resource folk the only way I have found is to open the program,App/Doc then resave it back.
Sometime if you pull the files to the Mac Desktop and copy them back will work. (Heavly dependant on extension loaded up on the Mac doing the work)

Pulling the Matrix cable is only a shot in the foot - going from a 1 gig link to a 100 port. But sometimes you got to do what you have to do

Hope this helps also....

 

[SteelBurner]

Checked the VLAN settings - there are none (yet).

I know that pulling the matrix cable isn't that brilliant, but imagine around 35 Mac/PC users trowing coffee cups and stuff like that after you... -o)

Anyway, didn't had the time to regroup the ports but will do it within the next days.

Thank you AlphaZ,

SteelBurner

 

[fbraga]

I have a similar problem. I have 2 3Com 3300 stacked with a matrix cable and 4 other switches like 3Com Stack Switch 610 connected to the second stack switch via a simple CAT5-e cable (100Mbs).
The first stack switch is used to servers, the second one is used to advanced users and interconnect others switches and the finals 610 switches are used to connect simple users to network.
Sometimes the second stacked switches blinks ports lights, like a flood and lost the comunication with the first switch.
My 3300's switches are running firmware 2.60 and i've tried to change some config parameters, like forwarding mode to store and forward, i've disabled broadcast storm, spanning tree, FastIP, VLAN and Multicast learning; reduced ageing time to 300s.
I'm thinking to upgrade the firmware to version 2.70.

Does anyone experienced this case or something like this?

Thanksful.

 

[SteelBurner]

fbraga,
I started this thread because of the problems you also experienced. I had an earlier version of the firmware on the switches and upgraded to 2.70 but still havin' the same troubles.

SteelBurner