[plug] NBN failure mode?

William Kenworthy billk at iinet.net.au
Fri May 8 10:18:07 AWST 2020


Thanks Byron,

    I have had really good results with Realtek up until now - but I am
aware some products have "issues" under linux.  In this case it created
quite confusing symptoms, no alarms and no errors or misconfiguration
that can be seen.

In this case, the dev involved decided to take advantage of the larger
buffer available on the r8153 chipset by splitting the RX and TX buffers
and changing the RX  buffer size to 32kb if an r8153 is detected - this
works fine with USB3, but has a problem with USB2.  I have changed the
buffer size back to 16kb in the kernel source (5.4.38) and recompiled it
- both of the devices I have are now running fine on this kernel at the
moment with one device connected to the NBN and one as part of a bonded
pair (the other part of the pair is a different, internal RTL chip on a
PCI bus) to an internal switch each running at 1Gbs.

BillK


On 8/5/20 9:32 am, Byron Hammond wrote:
> Hi Bill
>
> I too have run into Realtek driver problems.
> My work laptop had a Realtek Ethernet card.
> I tried installing Ubuntu 19.04 and it would lock up during install
>
> My solution was to blacklist the driver and use wifi.
>
> I believe the problem was fixed in a later Ubuntu release though
>
> Can't remember the chipset or driver though.
> 8163?
>
> from my mobile
>
> On Thu, 7 May 2020, 8:00 pm William Kenworthy, <billk at iinet.net.au
> <mailto:billk at iinet.net.au>> wrote:
>
>     The problem has now been resolved at my end by downgrading to a
>     4.19.120 kernel:
>
>        Speedtest by Ookla
>
>          Server: IX Australia - Perth (id = 1931)
>             ISP: iiNet Limited
>         Latency:     2.86 ms   (0.06 ms jitter)
>        Download:    *94.54 Mbps* (data used: 71.2 MB)                               
>          Upload:    *37.94 Mbps* (data used: 17.1 MB)                               
>     Packet Loss: Not available.
>      Result URL: https://www.speedtest.net/result/c/9651f666-4450-4363-972c-91cfeb20cf8f
>
>     The original speedtest using a number of the 5.4 series kernel
>     versions was (it did not matter what server was used):
>
>        Speedtest by Ookla
>
>          Server: AARNet - Perth (id = 6153)
>             ISP: iiNet Limited
>         Latency:     2.84 ms   (0.14 ms jitter)
>        Download:    *23.11 Mbps* (data used: 16.2 MB)                               
>          Upload:    *28.00 Mbps* (data used: 49.7 MB)                               
>     Packet Loss:     0.0%
>      Result URL: https://www.speedtest.net/result/c/4edaa6e9-167e-49f1-9f86-42dae678fce5
>
>     The problem - a patch
>     (https://bugzilla.kernel.org/show_bug.cgi?id=205923) added to the
>     linux kernel r8152 driver for the 8153 chipset in early 5.4 and
>     up.  I believe bug is still present in current (vanilla) kernel
>     versions.
>
>     Its specific to the Realtek RTL8153 USB chipset - my bad luck I
>     have two of them from different manufacturers :(
>
>     It would be interesting if anyone else has tripped over this as well?
>
>     Bill K.
>
>
>     On 6/5/20 9:02 am, William Kenworthy wrote:
>>     Thanks all, it was more a winge by me about using using unskilled
>>     helpdesk staff who follow a windows based script and little
>>     understanding if you cant use the very specific tools they want.  What I
>>     was trying to ask them "can you see any problem at the NBN interface"
>>     similar to what they used to do in the ADSL days.
>>
>>     I am now looking into if its the USB NIC (RTL8153 - I have two now,
>>     different brands but same results) to the NBN box and possibly a kernel
>>     5.4 driver problem but I am hamstrung by time (I need it up for those
>>     working from home) and suitable hardware.
>>
>>     BillK
>>
>>
>>     On 6/5/20 8:44 am, Gregory Orange wrote:
>>>     Hi Dean,
>>>
>>>     Thank you for posting this detail, and for humbly noting your credentials. Having a decent source of this sort of knowledge is very useful since I, like many PLUGgers I'm sure, am the "expert on everything with a plug" in my communities. I spend a lot of words on "that's not my area" but I do like to be able to follow up with "someone whose area it *is* once said..."
>>>
>>>     Cheers,
>>>     Greg.
>>>
>>>
>>>     On 5/5/20 10:28 pm, Dean Bergin wrote:
>>>>     Hello William,
>>>>
>>>>     As a network engineer, I deal with nbn issues on an almost daily basis so it's no surprise and somewhat justified for an RSP
>>>>     to ask for you to troubleshoot your end (albeit annoying), but seemingly unfair to ask you to use some abitrary windows-based application. Speedtest.net (web version) and/or iperf should suffice in this case.
>>>>
>>>>     At this point you though (and assuming you are a residential customer), you have a few options available to you;
>>>>
>>>>     1. You could comply and try run their application using an evaluation copy of windows. If ethernet is still a problem, then consider a USB-Ethernet adaptor. All up it wouldn't cost you much except a certain time investment. Heck you could probably do it in a VM on your Linux distro and avoid a bare-metal windows install and trash the thing when you're done with it.
>>>>
>>>>     2. You could ignore the RSP's request and demand that the matter be escalated, but the RSP may be within their rights to charge you, if the cause or fault is not with nbn equipment or RSP's backhaul etc. For example your equipment is faulty (could simply be a faulty lead) or perhaps misconfigured.
>>>>
>>>>     3. You might choose to move to another RSP that offers better support and potentially better service - The threat of a customer leaving can sometimes also put pressure on an RSP to fix a fault even if it's not theirs but you risk being without service should customer retention fail on their part...
>>>>
>>>>
>>>>     Kind Regards,
>>>>
>>>>     /Dean Bergin/.
>>>>
>>>>
>>>>     On Tue, 5 May 2020, 13:31 William Kenworthy, <billk at iinet.net.au <mailto:billk at iinet.net.au> <mailto:billk at iinet.net.au> <mailto:billk at iinet.net.au>> wrote:
>>>>
>>>>         Has anyone seen a failure mode for NBN connections where on a 100/40
>>>>         account it is only getting ~23/33 (down/up) using speedtests?  I am
>>>>         hoping to locate to an NBN or ISP fault.
>>>>
>>>>         The account is iinet and and trying to work through their overseas
>>>>         support drones is really frustrating ("plug your laptop directly in the
>>>>         NBN box" - I dont have a laptop with an ethernet connector", "download
>>>>         this windows only test software" - seriously, I don't have a windows
>>>>         machine ...)
>>>>
>>>>         Bill K.
>>>>
>>>>
>>>>         _______________________________________________
>>>>         PLUG discussion list: plug at plug.org.au <mailto:plug at plug.org.au> <mailto:plug at plug.org.au> <mailto:plug at plug.org.au>
>>>>         http://lists.plug.org.au/mailman/listinfo/plug
>>>>         Committee e-mail: committee at plug.org.au <mailto:committee at plug.org.au> <mailto:committee at plug.org.au> <mailto:committee at plug.org.au>
>>>>         PLUG Membership: http://www.plug.org.au/membership
>>>>
>     _______________________________________________
>     PLUG discussion list: plug at plug.org.au <mailto:plug at plug.org.au>
>     http://lists.plug.org.au/mailman/listinfo/plug
>     Committee e-mail: committee at plug.org.au <mailto:committee at plug.org.au>
>     PLUG Membership: http://www.plug.org.au/membership
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.plug.org.au/pipermail/plug/attachments/20200508/57345ff6/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pEpkey.asc
Type: application/pgp-keys
Size: 2188 bytes
Desc: not available
URL: <http://lists.plug.org.au/pipermail/plug/attachments/20200508/57345ff6/attachment.key>


More information about the plug mailing list