Roaming causes database to drop connection

Discussion in 'Wireless Networking' started by Dave Ploch, Nov 11, 2004.

  1. Dave Ploch

    Dave Ploch Guest

    We have recently expanded our site and now need to add 2 more access
    points. The first was an Aironet 1100 (Cisco). We have purchased two
    more of these.

    With all three ap's set to the same SSID / different channels I have a
    tablet (Toshiba running Tablet XP w/ Intel 2200b/g wireless) and roam
    around the area doing production management functions using an app
    (purchased) connecting to an MSSQL db in the office. Everything seems
    to work fine, it roams to each of the AP's but every once in a while it
    causes the app to lose communications with the db. It appears that this
    happens if the system decides to roam while a query is being processed.

    At this point the only authentication being done is MAC address
    filtering and no encryption. (starting out as easy as possible since
    the security issues are fairly limited in this installation).

    I read the thread going on about WDS and have read the tech notes on the
    Cisco site but this looks quite complex and I don't want to start down
    the wrong path since it looks like a significant learning curve.

    Any ideas on what we can do to resolve this issue?

    Dave Ploch, Nov 11, 2004
    1. Advertisements

  2. Assuming everything is set up correctly with the AP's (they all have the
    same SSID, the same security configuration, and user the same DHCP server),
    it's still possible that you're losing your wireless connection. This would
    be a function of the wireless driver, not XP. Ideally the driver will switch
    between access points seamlessly (without disconnecting). XP only plumbs
    down the SSID and the security settings.

    That said, you can try an experiment to check your connectivity: walk around
    running "ping <ipaddress_of_database_server> -t -w 1". There may be some
    "dead" spaces in your network coverage that would cause the driver to
    disassociate from the AP. You can optionally experiment with ping's length
    value (see ping -? for help) to see if the problem only happens with larger
    packet sizes.

    You also might want to do an "ipconfig /all" after you lose a connection -
    to make sure you have the same IP (all AP's should be served by the same
    DHCP server so you get the same IP address).

    If you actually ARE losing connectivity (meaning that the network
    disconnects), then you won't be able to send or receive packets, the media
    state of your adapter may change, etc. Ideally your application would be
    robust enough to accomodate brief network outages, but if this is not the
    case, you can try using a static IP address (and excluding that address from
    the DHCP lease pool).

    If you do know for sure that you are losing connectivity, make sure your
    network is at the top of the list in the wireless settings and the SSID is
    NOT hidden on the AP's. Some wireless drivers have a tough time with hidden
    SSID's and Windows only looks for them when all visible networks have been
    ruled out as possible connections.

    If you try to use a static address, as a last resort you might also want to
    try changing the DisableDHCPMediaSense registry value. I haven't personally
    tried this, but it might help your laptop think the connection is still
    active (even when it is not). Only try this with a static address (not
    DHCP). See . You may need to
    reboot for it to take effect.

    You are likely using a TCP connection for your database access. If you want,
    you can try sniffing with netmon to see whether (and why) the database
    connection is closing. (Does it happen on the client side or server side?)
    Does the routing infrastructure you have in place take time to start
    shifting packets from one subnet to another? If each AP is on a different
    subnet then you're probably using different DHCP servers anyway and the TCP
    connection dropping would be a side-effect of an IP-address change.
    Carl DaVault [MSFT], Nov 18, 2004
    1. Advertisements

  3. Also, since this seems to work "sometimes" for you, if you have a different
    wireless card, you can disable the internal one and see if the new just one
    roams better.

    Always use the "latest" drivers from your NIC manufacturer. Some older ones
    can be buggy. : )

    Standard Disclaimers -
    This posting is provided "AS IS" with no warranties,
    and confers no rights. Please do not send e-mail directly
    to this alias. This alias is for newsgroup purposes only.
    Carl DaVault [MSFT], Nov 18, 2004
  4. Dave Ploch

    Dave Ploch Guest


    Thanks for the ideas.

    I have already done a site survey using continuous pings to find dead
    spots. I have significant overlap from the three AP's and did not find
    any dead zones.

    We DO have the most current 2200bg drivers for the Toshiba tablet. The
    three Cisco Aironet 1100's have the most current IOS.

    The SSID's are not being transmitted. I'll try turning those back on
    for a test.

    As I said earlier, it looks like the connection to the database (sybase)
    fails during a roam event but ONLY when it is the midst of
    communications. If the roam event happens in between queries to the db,
    everything works fine and the next query works great.

    Dave Ploch, Nov 18, 2004
  5. Dave Ploch

    Dave Ploch Guest

    I also have a Cisco card and will try in in the tablet to see if that
    works any better.

    Thanks for the help,

    Dave Ploch, Nov 18, 2004
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.