kennysarmy Posted June 28, 2010 Posted June 28, 2010 I have a core switch 3 com model : 3Com Switch 5500G-EI 48-Port Software Version 3Com OS V3.03.02s56 It has connected to it via UTP two 3Com Baseline Switch 2948 Plus's These are connected to the core switch via two UTP cables trunked for extra bandwidth....I've had nothing but problems with clients doing GPO updates connected to the 2948's so decided to do some testing with a spare Allied Telesis switch. I get exactly the same issues with the AT switch as I do the 3 com switch. See attached screenshot.... If I connect the clients direct to the core switch (5500G) then all works fine every time. See attached screenshot I thouhgt initially the issue was with the 3 com edge switches but since the AT edge switch does the same now I'm not convinced. My next trial is going to be to connect the AT switch to the core not using copper but using a fibre lead and SPF modules. If anyone else can offer any suggestions... How could I test the network speeds from the clients as they are now going through the AT switch before I uplink it with copper?
mavhc Posted June 28, 2010 Posted June 28, 2010 iperf, windows compiled version is older than the latest
djm968 Posted June 28, 2010 Posted June 28, 2010 I've had some good results with this free tool. Qcheck - Free Network Benchmark Utility » IxChariot™ - Test Device and System Performance under Real-World Load Conditions
kennysarmy Posted June 28, 2010 Author Posted June 28, 2010 Thanks. Current network speed between client and server using iperf is: 93.2 Mbits/sec That seems OK to me....need to retest when switch is connected via fibre.
keithu Posted June 28, 2010 Posted June 28, 2010 I would check duplex settings on the link between the core and edge switches. Also make sure the trunked connection is actually configured as a trunk. Have you tried it with a single copper link instead of the trunked pair?
kennysarmy Posted June 29, 2010 Author Posted June 29, 2010 (edited) I would check duplex settings on the link between the core and edge switches. Also make sure the trunked connection is actually configured as a trunk. Have you tried it with a single copper link instead of the trunked pair? Thanks for your suggestion: The edge switch is trunked: Port State Enabled PVID 1 Flow Control Disabled Speed Auto[1000M] Duplex Auto[Full] This connects to the core switch which settings are: Port State Enable[Active] PVID 1 Flow Control Disable Link Type Trunk MDI Auto Speed Auto[1000] Duplex Auto[Full] Max MAC Count No Limit Jumbo Frame Disabled Port Isolate Disabled Broadcast Suppression 3000(pps) Edited June 29, 2010 by kennysarmy
Guest Guest Posted June 29, 2010 Posted June 29, 2010 Eliminate the trunk first, sounds to me like that might be the issue.
kennysarmy Posted June 29, 2010 Author Posted June 29, 2010 I have attached the AT switch using SFP modules and a fibre patch lead to one of my 3 com fibre core distribution switches, these are: 3Com Switch 5500G-EI SFP 24-Port Software Version 3Com OS V3.03.02s56 I am going to do some testing with GPO updates today on some test PC's and see if this works OK.
kennysarmy Posted June 30, 2010 Author Posted June 30, 2010 UPDATE: Connecting the AT (edge) switch to the CORE via spf modules/fibre patch leads : WORKS. I did think this might isolate the problem as suggested above to the PORT TRUNKING - however, when we tested the AT switch connected to the 3-com CORE switch 5500G-EI 48 by just a single UTP lead we STILL HAD PROBLEMS WITH GPO updates.... Any suggestions as to how to proceed? Regards.
SYNACK Posted June 30, 2010 Posted June 30, 2010 I don't think that this is network speed related. This kind of error usually happens when the network card takes to long to get an IP address and Windows gives up waiting and plows on without it. With XP there was an old disable mediasense trick that made it wait until it had an IP before proceeding but Vista+ are far more robust and have never needed kind of fix in most situations. I think that you problem could be to do with STP (spanning tree protocol) being enabled on the prts of your switch used by the client computers, this can put the port into a lock down for around a minute with no access as a safety measure against network loops. If it is on in the edge switchs there should be a client port mode for STP that only isolates for 5 seconds which should let the system pick up the IP address quickly enough and still provide a bit of protection. If this is not the cause I would look at using something like Wireshark to see what is going on directly on the network and see if it is the machines not being able to send out DHCP requests eairly enough or if it is simply not getting a quick enough responce from the DHCP server. 1
kennysarmy Posted June 30, 2010 Author Posted June 30, 2010 Thanks. I have researched further on the web and found this snippet....is it accurate ? Spanning tree should be enabled on the links between switches, and disabled on those to the servers.
SYNACK Posted June 30, 2010 Posted June 30, 2010 This depends on who you listen to, STP should be setup on links between stitches and should generally be set to client port mode on all other ports. It should be safe to disable it on your server ports as those are unlikely to be looped back and will be the first ones you check anyway. For testing I would just disable it on the client ports of a few machines experiencing the issue and see if that resolves it, if it does I would still look to put it into client port STP mode which should still work but afford you some extra protection. Using STP allows for multiple redundant links between switches and also helps prevent broadcast storms which will bring your network to a standstill if there is a loopback. If you are using STP though you do need to have the core switch setup with the highest priority when using multiple redundant links otherwise it can lead to some very strange traffic flow. To work the network does not specificly need STP enabled, it just gives you the extra features and protections meantioned above.
spc-rocket Posted June 30, 2010 Posted June 30, 2010 (edited) Yes that is correct, ideally you should disable the STP on the access ports i.e. ports that connects to workstations, servers, printers or any other end device. It should be enabled for switch to switch ports. in Cisco this is called portfast which put the port into forwarding mode immediatly and does not wait for the loop detection process to complete. Ash. Edited June 30, 2010 by spc-rocket spelling correction 1
kennysarmy Posted June 30, 2010 Author Posted June 30, 2010 3-com : edge switch 1 looks like this when I check the GUI under spanning tree... Port Status Path Cost Edge Port State Link Type Port Priority 1 Enabled 200000 Disabled Discarding Shared 128 2 Enabled 200000 Disabled Discarding Shared 128 3 Enabled 200000 Disabled Discarding Shared 128 4 Enabled 200000 Disabled Discarding Shared 128 5 Enabled 200000 Disabled Discarding Shared 128 6 Enabled 200000 Disabled Discarding Shared 128 7 Enabled 200000 Disabled Discarding Shared 128 8 Enabled 200000 Disabled Discarding Shared 128 9 Enabled 200000 Disabled Discarding Shared 128 10 Enabled 200000 Disabled Discarding Shared 128 11 Enabled 200000 Disabled Discarding Shared 128 12 Enabled 200000 Disabled Discarding Shared 128 13 Enabled 200000 Disabled Discarding Shared 128 14 Enabled 2000000 Disabled Forwarding Point to Point 128 15 Enabled 200000 Disabled Discarding Shared 128 16 Enabled 200000 Disabled Discarding Shared 128 17 Enabled 200000 Disabled Discarding Shared 128 18 Enabled 2000000 Disabled Forwarding Point to Point 128 19 Enabled 200000 Disabled Forwarding Point to Point 128 20 Enabled 200000 Disabled Discarding Shared 128 21 Enabled 200000 Disabled Discarding Shared 128 22 Enabled 200000 Disabled Discarding Shared 128 23 Enabled 2000000 Disabled Forwarding Point to Point 128 24 Enabled 10000 Disabled Forwarding Point to Point 128 25 Enabled 20000 Disabled Forwarding Point to Point 128 26 Enabled 200000 Disabled Discarding Shared 128 27 Enabled 200000 Disabled Discarding Shared 128 28 Enabled 200000 Disabled Discarding Shared 128 29 Enabled 200000 Disabled Forwarding Point to Point 128 30 Enabled 200000 Disabled Discarding Shared 128 31 Enabled 200000 Disabled Discarding Shared 128 32 Enabled 200000 Disabled Discarding Shared 128 33 Enabled 200000 Disabled Discarding Shared 128 34 Enabled 200000 Disabled Discarding Shared 128 35 Enabled 200000 Disabled Discarding Shared 128 36 Enabled 200000 Disabled Discarding Shared 128 37 Enabled 200000 Disabled Forwarding Point to Point 128 38 Enabled 2000000 Disabled Forwarding Point to Point 128 39 Enabled 200000 Disabled Discarding Shared 128 40 Enabled 200000 Disabled Forwarding Point to Point 128 41 Enabled 200000 Disabled Discarding Shared 128 42 Enabled 20000 Disabled Forwarding Point to Point 128 43 Enabled 200000 Disabled Forwarding Point to Point 128 44 Enabled 200000 Disabled Discarding Shared 128 45 Enabled 20000 Disabled Discarding Point to Point 128 46 Enabled 20000 Disabled Forwarding Point to Point 128 47 Enabled 20000 Disabled Discarding Point to Point 128 48 Enabled 10000 Disabled Forwarding Point to Point 128 Ports 24 & 48 are the trunk to the 5500 core switch... Anyone know if this looks right or wrongly setup? Cheers.
kennysarmy Posted June 30, 2010 Author Posted June 30, 2010 Amended and now looks like this: 1 Disabled - - - - - 2 Disabled - - - - - 3 Disabled - - - - - 4 Disabled - - - - - 5 Disabled - - - - - 6 Disabled - - - - - 7 Disabled - - - - - 8 Disabled - - - - - 9 Disabled - - - - - 10 Disabled - - - - - 11 Disabled - - - - - 12 Disabled - - - - - 13 Disabled - - - - - 14 Disabled - - - - - 15 Disabled - - - - - 16 Disabled - - - - - 17 Disabled - - - - - 18 Disabled - - - - - 19 Disabled - - - - - 20 Disabled - - - - - 21 Disabled - - - - - 22 Disabled - - - - - 23 Disabled - - - - - 24 Enabled 10000 Disabled Forwarding Point to Point 128 25 Disabled - - - - - 26 Disabled - - - - - 27 Disabled - - - - - 28 Disabled - - - - - 29 Disabled - - - - - 30 Disabled - - - - - 31 Disabled - - - - - 32 Disabled - - - - - 33 Disabled - - - - - 34 Disabled - - - - - 35 Disabled - - - - - 36 Disabled - - - - - 37 Disabled - - - - - 38 Disabled - - - - - 39 Disabled - - - - - 40 Disabled - - - - - 41 Disabled - - - - - 42 Disabled - - - - - 43 Disabled - - - - - 44 Disabled - - - - - 45 Disabled - - - - - 46 Disabled - - - - - 47 Disabled - - - - - 48 Enabled 10000 Disabled Forwarding Point to Point 128 going to test GPO updates via this switch shortly...
kennysarmy Posted June 30, 2010 Author Posted June 30, 2010 WHHHHHEEEEEYYYYYYYYY All back to normal..... Thanks guys.... Thanks left
sted Posted June 30, 2010 Posted June 30, 2010 when you say gpos arnt applying is it the computer policies at first bootup? i have in the past had issues with dell gx something old (intel pro 1000 iirc) cards and hp switches whereby if i set the pc card to auto it wouldnt get gpos set to 10mb 1/2 (or possibly 100mb 1/2) policys applied correctly flashed the switch and put pcs back to normal all gpos applied also the reg file below has been known to help Windows Registry Editor Version 5.00 [HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon] "GpNetworkStartTimeoutPolicyValue"=dword:0000003c
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now