none
Network routing issue.

    Question

  • Hi all.

    Hopefully you guys can help. I have a virtual server running a application for a web-based storefront. It has been running really slow lately. At times i get time-out errors. The web based side of things seem to run fine.

    Only when our server is accessed does the slow speeds occur. The computers accessing the server are on a network of 120 pc's with a bunch or routers and switched throughout the property. I have plugged a pc running the software into ANY port on the same switch that is connected to the server and i get TURBO speeds. Any other switch or router anywhere on the network i get lag and occasional time-outs. I do maintenance on the server to speed up indexing but it cant be this as it runs really fast when connected to the above mentioned switch.

    From what i can gather its some sort of routing issue. There seems to be some confusion or something. Seems like the routing table or ARP refresh or something to that nature. Firewall perhaps? I don't have tons of Networking EXP and my boss is out of town. Would be nice to sort this out. This is all on the same domain.

    Any help is greatly appreciated!

    Thanks

    Chris

    update: I just learned  that even though it is getting turbo speeds connected to the one switch it still is directed via web. So the same pc was tested on two different switches i have in our closet and one switch is super fast one is the same as anywhere else on the network. IIS  .NEt issue SQL??

    • Edited by chrisspliff Wednesday, September 11, 2013 11:30 PM
    Wednesday, September 11, 2013 11:07 PM

All replies

  • Hi,

    In order to troubleshoot the issue, you can run command: tracert/pathping“server IP address”.

    You can see how the packets arrive the server and how much time it costs.

    Since you have only 120 clients, I don’t think you need many routers.

    Simplify your network maybe a good choice.

    Can you provide the topology of your network if you need further support?

    Hope this helps.

    Friday, September 13, 2013 6:05 AM
  • I tried the pingpath command and am coming back with a 0MS latency. Looks like everything is running perfectly fast with an exception of a few *.apsx files found on the server. Apparently the software is touchy and can be easily made inefficient from coding adjustments. I would have accepted the fact that it just runs slow but since I noticed immediate load times once plugged into the main switch connected to the server, this leads me to believe otherwise. ... let me do some more diagnostics.

    Thanks for the help.

    Monday, September 16, 2013 4:00 PM
  •  but since I noticed immediate load times once plugged into the main switch connected to the server, this leads me to believe otherwise.

    how is your network topology looks like?

    When you said you plugged to the main switch and works fine, basically what you are saying is you have a lot of cascade network switches?

    And it works fine when the machine that hosts the site is connected  to a switch which the server is also connected?

    Like:

    Internet<--Router-->Internal Network-->Switch 1-->Server connection

    Any VLAN configuration or hardware firewall on your environment?

                   


    Every second counts..make use of it. Disclaimer: This posting is provided AS IS with no warranties or guarantees and confers no rights.

    Wednesday, September 18, 2013 8:06 AM
  • What is your hypervisor? Virtual OS ? Have you multiple networks?

    Maybe there is something to do with the network cards and the OS, like offload...

    You can have a look here how to disable offloading (on the vOS):

    http://technet.microsoft.com/en-us/library/cc959732.aspx

    http://support.microsoft.com/kb/951037

    Don't forget to backup registry before....

    Wednesday, September 18, 2013 8:38 AM
  • I have my main Fiber line running into a switch in a nearby building. This Fiber is then connected to switch number 2 here in my main server closet. Switch 2 is connected to switch 1(also here in my closet..there are 2 switches total in my closet) the dedicated shop server is directly connected to switch 1. I decided to grab a PC located in a completely different building and connected it as close to the server as possible to test in order to eliminate faulty wiring or routing issues etc..

    From this I found switch 1 lets the PC access the server with no lag whatsoever, switch 2 is slow. Anywhere else on my network i get the same predictable slow speeds.

    There are about 20 switches located amongst 15 buildings throughout our property. Like I said the Main line comes into a switch then connected to my main closet switch here so I'm thinking I am bypassing a bunch of stuff to simplify things. Could there possibly be upload speed restrictions in place?

    Specs:

    Single Network.

    We have one Barracuda Spam and Virus Firewall 200 on this network.

    The switches are Linksys SRW 2048's.

    Our shop server is on a dedicated HPproliant DL120 G6.

    No VLAN config but im not certain.. we run VMware, server 2003 OS R2 for virtual machines. (my understanding is its a dedicated server but it looks like it is running VMware to run our web based shop)

    I will try to post a diagram soon..

    Thanks for helping!



    • Edited by chrisspliff Wednesday, September 18, 2013 3:32 PM
    Wednesday, September 18, 2013 3:25 PM
  • Ludovic

    See above post thanks

    Wednesday, September 18, 2013 3:26 PM
  • ok, if you have more that one VM try that to the web based shop and to the other VM and to VMware Server (on the same VMware server):

    ping SERVER_IP -l 1400 -f -n 50

    (it will do 50 ping query and display stats after)

    Do that also from the laptop and on the two switches.

    Next: if you can monitor your two switches, try  to see the stats of the port between the two switches (I see that you have Linksys, just go on the web interface to have the port stat) then look if there is something abnormal: collison, packet drop...

    Next:You can also ping switch 1 when you are connected to switch 2 with your laptop with "ping Switch2_IP -l 1400 -f -n 50"

    Read the part about disable taskoffload it's also true with Windows 2003:

    http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1009517

    HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters Value(DWORD): DisableTaskOffload = 1

    Post your test result when you have it.

    Wednesday, September 18, 2013 3:55 PM
  • FYI:

    This is what I set on my VM in 2003 server (but I use subnets, about 60 subnets):

    Windows Registry Editor Version 5.00

    [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters]
    "DisableTaskOffload"=dword:00000001
    "EnableTCPChimney"=dword:00000000
    "EnableTCPA"=dword:00000000
    "EnableRSS"=dword:00000000

    Wednesday, September 18, 2013 4:11 PM
  • So i have been pinging my devices and found out this:

    We have 2 switches in our closet.

    Switch 1 and switch 2 both ping from anywhere in the network.

    When directly plugged into switch 1 I can only ping switch 2 not switch 1, also when directly plugged into switch 2 i can only ping switch 2 not switch 1. So looks like switch 1 cannot be pinged

    Is there a block or something here?

    Switch 1 is the switch that produces normal speeds and switch 2 produces the lag and occasional timeouts.

    Also i just connected another computer that has not been set up for using our web based software and it is able to ping both switches.(same cable location onto switch)



    • Edited by chrisspliff Thursday, September 19, 2013 6:58 PM
    Thursday, September 19, 2013 6:52 PM

  • When directly plugged into switch 1 I can only ping switch 2 not switch 1, also when directly plugged into switch 2 i can only ping switch 2 not switch 1. So looks like switch 1 cannot be pinged



    most probably those are manage switches, if those are manage switches network administrator can set to block ICMP from either switches.

    Can you confirm whether switch1 is being trunk to switch2? What i mean is there a cable from switch1 connecting to switch2.

    And if you have a laptop on switch1, and you have a laptop on switch2. If you share a file (just for testing) is the shared file can be seen on either switches?

    if the shared file can't be accessed from either switches, then there's configuration on the switch that is set to block network traffic.


    Every second counts..make use of it. Disclaimer: This posting is provided AS IS with no warranties or guarantees and confers no rights.

    Friday, September 20, 2013 1:18 AM
  • So, for the ping:

    Test laptop <---> switch 1 <---> switch 2 OK

    Test laptop <---> switch 1 NOK

    Simple computer <--> switch 1 (same port) OK

    Test laptop <---> switch 2 <---> switch 1 NOK

    Test laptop <---> switch 2 OK

    Correct ?

    Can Do that and copy/past it here: 

    For each test do this sequence:

    before each test do,

    ipconfig /all

    for each ip do,

    ping IP -l 1400 -f

    ping IP

    tracert IP

    Test 1:

    laptop on switch 1

    do the sequence with IP of switch 2, Virtual machine.

    Test 2:

    laptop on switch 2

    do the sequence with IP of switch1, switch 2, Virtual machine.

    Test 3:

    Simple computer (on switch 2):

    do the sequence IP of switch 1, IP of switch 2.

    And paste all the results as it is displayed.


    Friday, September 20, 2013 7:18 AM
  • Solved! :)

    Turns out one switch was running at 100MB/s and the other at gigabit speeds. So we patched the server into the gigabit switch and got good speeds.  We have fiber coming into the "gigabit" switch also the other "slow" switch is daisy chained to  the "gigabit"switch. How come they are not both running at gigabit speeds? They are both the same model switch. Why is one switch only running at 100mb? Just would like to fully understand how the routing is causing one switch to not run at full speed.

    Thanks for everyone's help!

    Tuesday, September 24, 2013 10:20 PM
  •  How come they are not both running at gigabit speeds? They are both the same model switch. Why is one switch only running at 100mb? Just would like to fully understand how the routing is causing one switch to not run at full speed.

    Thanks for everyone's help!

    you can limit the bandwidth on the interface, if someone has done it. (if you are authorized to login to your switch and if it is a Cisco router, you can use "show run" to view the configuration of your switch).

    if no one had configured the bandwidth interface then the switch is faulty already.

    search Bing, with this keyword "cisco switch limit bandwidth" 

    hope it helps.


    Every second counts..make use of it. Disclaimer: This posting is provided AS IS with no warranties or guarantees and confers no rights.


    • Edited by cguan Wednesday, September 25, 2013 6:47 AM edit
    Wednesday, September 25, 2013 6:47 AM

  • Besides the router being restricted due to setup or being faulty, Is there any way the cabling (daisy chain) between the switches could cause the slow speeds? any certain port i should use to connect the two switches together so both run at full speed.

    Thursday, September 26, 2013 10:27 PM
  • life would be easier if you can type "show run" (if it is a Cisco router) it will tell you everything.

    Which interface is configured with this and that, rather than guessing.

    If there is a faulty cable, it will not have any connection at all or either it will be an intermittent network connectivity failure.

    If you are not able to tell the running configuration of the switches, it will be very very difficult to troubleshoot. (if it is a manage switch, if it is not a manage switch and it's causing this issue then I guess the switch could be faulty)


    Every second counts..make use of it. Disclaimer: This posting is provided AS IS with no warranties or guarantees and confers no rights.


    • Edited by cguan Friday, September 27, 2013 3:23 AM edit
    Friday, September 27, 2013 3:21 AM