They are the best Ethernet trolls around.

Josh Luthman
Office: 937-552-2340
Direct: 937-552-2343
1100 Wayne St
Suite 1337
Troy, OH 45373
On Jan 27, 2015 7:42 PM, "Forrest Christian (List Account)" <
[email protected]> wrote:

> The most ironic thing about my whole rant issue - and I just realized it
> late last night.
>
> When I finally *want* a 100Mb/s link, it gives me a 10Mb/s one, and
> refuses to play nice at 100Mb/s - only 10.
>
> -forrest
>
> On Tue, Jan 27, 2015 at 5:09 PM, George Skorup (Cyber Broadcasting) <
> [email protected]> wrote:
>
>> That Base-II you replaced for me made this quite obvious. It would not
>> work in any port unless forced to 10HD. That's the one I did all of the
>> polarity swapping on and nothing would make it work with auto-nego. Worked
>> perfectly fine plugged into any old dumb or managed switch. Nothing
>> MikroTik worked. Well besides forcing 10 half.
>>
>>
>> On 1/26/2015 4:19 PM, Forrest Christian (List Account) wrote:
>>
>>> As many of you have heard, I've had a bit of a irritation for quite some
>>> time about Mikrotik's seeming inability to consistently wire their Ethernet
>>> ports correctly for 10mb/s operation - specifically paying attention to the
>>> correct *polarity* of the wires in the Ethernet pairs.  This wreaks havoc
>>> with devices which need the correct polarity to work, such as my
>>> SiteMonitor Base II's.   For instance, every other switchport on the
>>> RB250GS Ethernet switch doesn't work.   I don't remember if it's 1 3 and 5
>>> which doesn't work, or if it's 2 and 4, but it is definitely every other
>>> one.   Swapping the wires in one of the pairs fixes the problem, so I'm
>>> sure that's the cause.
>>>
>>> I've tried over the years to get MikroTik to care enough to even
>>> acknowledge the issue.   Of course, the problem isn't their lack of
>>> compliance with the standard, but instead it must be my problem instead, so
>>> I've resolved to just dealing with the occasional need to train a customer
>>> to try a different port.
>>>
>>> I should note that as far as I can tell, this isn't a common problem
>>> with other switch vendors.   In fact, I don't think I have a single
>>> non-MikroTik device in my possession which doesn't work with the
>>> standard-complying 10Mb/s Ethernet port in the base unit - and I have quite
>>> a variety that I've tried.  Not sure how MikroTik is so good at getting
>>> this not-very-consistently wrong, and everyone else seems to almost always
>>> get it right.
>>>
>>> But, that's not really my rant.  It just provides the background.
>>>
>>> Over the last year or so I've slowly put together the internal network
>>> at home and at PacketFlux out of MikroTik devices.  Quite frankly, they
>>> have a feature set I appreciate, and I can generally overlook the issues,
>>> but they're starting to get on my nerves, the most recent ones being:
>>>
>>> * Routerboard 850Gx2 doesn't show master/slave configuration options in
>>> webfig (ok this one is minor).
>>>
>>> * That same routerboard 850Gx2, won't link with a 10Mb/s device at all
>>> unless you turn *off* autonegotiation and hard code it 10Mbs.   Well,
>>> except when you try to replicate it later, in which case it works fine with
>>> autonegotiation on. Fine, be that way.   Well, until hours later it decides
>>> to not link with that exact same 10Mb/s device and you have to set it back
>>> to not negotiate.   Fortunately, hardcoding the speed seems to always work
>>> so I'm leaving it that way.
>>>
>>> * Disabling advertising speeds seems to be generally non-functional on
>>> some devices.   Want to negotiate at only 10/100, so turn off 1000M
>>> checkboxes?   Fine, except the damn thing still autonegotiates at 1000M.
>>>
>>> And what really set me off this weekend:
>>>
>>> Remember that RB250GS?   Well, I recently upgraded my main
>>> development/desktop computer (much faster now, thanks), but am still moving
>>> things off my old one.   So I want it up to be able to VNC into it.   In
>>> addition, I have a newly installed Linux machine here as well.  The
>>> physical hardware for both of these devices are Dell workstations from
>>> about the same time period.   They both have 10/100 Ethernet cards which
>>> have been autonegotiating at 100Mb/s with when plugged into a rb2011 (which
>>> has it's own unique issues as well).
>>>
>>> So this weekend, I decide to move these things off the top of the r&d
>>> bench to under it.   I decide to install that RB250GS under the bench as
>>> well to plug them into and give me a couple more spare ports, and so I get
>>> it plugged in and linked up to another MikroTik (amazingly this link works
>>> just fine), and then hook both dell machines (The Win7 one and the Linux
>>> one) to the RB250GS.
>>>
>>> A few hours later, I copy some stuff from the old Win7 box to my new
>>> one.   Many gigs of stuff to be specific - something like 40GB or
>>> thereabouts.    Windows is nicely telling me that it's going to take 'more
>>> than a day' and that I'm only transferring at 1MB/s (yeah 10Mb/s).   WTF?
>>>
>>> So I get digging.  The (@#*$ 250GS is only linking at 10Mb/s.  Check the
>>> other machine: 10Mb/s also.  Fine. Negotiation issue.   Go in, make sure
>>> that the port is enabled for 100Mb/s.   Yep.  Turn off auto, force it to
>>> 100Mbs - do the same on the computer end.   Links up at 100Mb/s, Yay!
>>>
>>> Go back, look at the transfer speed.   It's gone DOWN? How is that even
>>> possible?  Oh, yeah the fact that the thing is throwing errors at 100Mbs.
>>>  Swap cables.  Swap Ports. Still no go.   Finally grab the end of the cable
>>> which is plugged into the 250GS and plug it into a spare port on the
>>> RB850Gx2 and it works.  100Mb/s, auto negotiate, transfer speeds of 10Mb/s
>>> between the computers.  Things transfer just fine.
>>>
>>> And then... as a final straw, I get a response to a trouble ticket from
>>> a customer complaining that they can't get a link with a RB493AH running
>>> latish code on any port.  No matter what the setting for negotiation is.
>>>
>>> Why is this even close to acceptable for a manufacturer who should be
>>> able to catch this stuff?
>>>
>>> -forrest
>>>
>>
>>
>

Reply via email to