That's one of the things I was thinking. Ping works all the way up to 1500 bytes. At 1501 we get the fragmentation error. So that's working like we expect. Didn't try too many intermediate packets sizes. Just 50, 100, 500, 1000, 1500.

bp

On 10/30/2014 6:03 PM, Harold Bledsoe via Af wrote:
When you ran the ping tests, did you try all sizes from 56-1500 with no fragment flag set? I've seen a case where 2 random sizes didn't pass which resulted in some very bizarre behavior. ;-)

-Hal

On Thu, Oct 30, 2014 at 8:58 PM, Bill Prince via Af <[email protected] <mailto:[email protected]>> wrote:


    This is a bizarre set of symptoms, and I really don't know what is
    going on.  So I will articulate the facts, and maybe one of you
    can tell me what might be wrong.

We have a business subscriber that occupies several buildings. The buildings are separated by enough distance that we have to
    interconnect by means other than vanilla ethernet.

    Our service is delivered to their main office.  Our SM is
    installed there (PMP450), plus a Mikrotik router on ROS 5.26.  The
    Mikrotik manages 4 VLANs; 1 business VLAN, which is bridged to the
    main subnet in the main office.  The other 3 VLANs are guest
    VLANs; each on their own (private) subnet.

    All the computers, etc. work fine in the main office.

    The main office is connected to the "guest building" with a VDSL
    modem (~~ 800' phone line between buildings).  Not much occurs in
    the guest building; it has a couple of WiFi APs for the guests.

    In the guest building, we've installed an RB260GS switch. It
    divides the various ports out to 4 different VLANs.  A couple
ports are the "business VLAN", plus 3 different "guest VLANs". The SFP port on the RB260GS is used to connect to the "satellite office" another couple hundred yards beyond the guest building. The SFP port is on the business VLAN.

    At the satellite office, they have 2 computers.  Everything on the
    2 computers in the satellite office seems to work just fine.  Web
    browsing, streaming youtube, etc.

    However, when they run Outlook, "some" email doesn't go to the
    destination.  As far as we can tell, it gets to their off-site
    SMTP server (Globat), but some of it doesn't ever reach its
    destination. If they use their web-based email, the email works
    every time. Also, the POP part of the email works just like you'd
    expect.

    Today, we moved one of the computers back to the main office, and
    surprise, surprise, Outlook starts working just like it's supposed to.

    We've run extended ping tests between the satellite office and the
main office, and there is no break in the link. It seems solid. So where/how is the SMTP part of email breaking?

    What tests can I run to figure this out?


-- bp




--

Harold Bledsoe


Reply via email to