Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[les-70]

  I picked up an unused ex AAISP  XyXEL VMG1312 on ebay a while ago and have just got round to trying it.  I had a bad experience with a VMG8324 -gui lockups and high error rates so I have been hesitating with the VMG1312.  I have started it running today around 5pm with the same heavy sync speed cap that the usual HG612 has had for quite a while.  I hope this will allow a comparison unbiased by the XyXEL's higher attainable syncs.  It is running on MDWS.

  So far all has gone well with nothing annoying me other than the TBB BQM appearance which really renders it almost pointless to run.  On the VMG1312 this involved enabling a firewall access setting for ICMP from the BQM IP address to the router.  The results since about 5pm today can be seen in the attached image -  massive packet loss.  I recall the VMG8324 being fairly odd but not this bad.  I tried playing with QOS and other settings but without any efffect on the BQM.

 Is this normal for a VMG1312? and can any setting be used to improve things?

 A pingtest.net test gives perfect results and apart from the BQM nothing seems at all amiss.

 Below I will report on error rates c.f. an Hg612.

[Weaver]

Either this is a faulty modem, or there's a serious line fault / outage. You need to swap back to a known good modem temporarily, in order to test whether the line is working at all, and then possibly get a line fault looked into. Sorry that's a bit vague, and fairly obvious, but everything needs testing systematically.

[PhilipD]

Hi

We've seen this before on ZyXELs   They tend to be configured with a rather sensitive threshold for identifying a denial of service attacks, and it seems TTB packet frequency is just about on that threshold so you get that pattern.

See this topic here http://forum.kitz.co.uk/index.php?topic=15271.0, further down the thread some info was added about how to bypass the DoS detection for Thinkbroadbands IP address.  The other option is to turn Dos off.

On the ZyXEL VMG8934 which I had this issue with, they have changed the threshold in new versions of the firmware and DoS can remain on without being triggered so easily, and so this problem was solved on that model at least.

Regards

Phil

[Weaver]

Shows you what I know!
Nasty.

@philipd - thanks, glad somebody knows what they're talking about. :-)

[les-70]

  Thanks, your correct I notice the recommended AAISP settings of allowing all pings and disabling DOS fixes the issue completely. However I don't like either of these settings.  When ever I am open to all pings I seem to get multiple port scans each day and to me it does seem best to be more hidden.  I have tried and failed to fix things via the acl rules so I think IPTABLES via telnet is the last option before either not using BQM or allows using it and allowing all pings.

[les-70]

  I attach the latest state of things. After the red comb all night there is a block between roughly 8am to 9am with all pings allows and dos off - things are 100% Ok if you don't mind these settings. 
   Then since about 9:30 I  have pings blocked but with npr's iptable rule "iptables -I INPUT 1 -s 80.249.99.0/24 -p icmp --icmp-type echo-request -j ACCEPT" which works OK but not quite so perfect as now there are a bit more exaggerated yellow spikes above the green.
  It is disappointing the ZyXEl acl rules can't do the same job as the iptable rule.  My impression is that on most routers there would be no difference.

[npr]

The yellow grass on the graph is normal on my Zyxel vmg8924, guess it may be the same for the vmg1312.

Using the latest firmware on the vgm8924 this is how I allow pings from/to TBB's ping graph, don't know if the same will work on your model:

Firewall settings > DoS protection is enabled.
Remote mgmt > ICMP Trusted Domain ticked (enabled)
Remote mgmt > Trusted Domain > TBB server added (80.249.99.164)

[les-70]

  The yellow grass does seem too bad -- see below.  Now after one day I am happy with the VMG1312 setup as it is doing all that I need.  I am however NOT happy with its error rate.  I have exactly the same position as I had with my once owned VMG8324.   The average daily error second rate of the HG612 capped to 52Mb/s down is 14ES/day on fast path. The maximum number of ES per day over a 60 day period at ths speed has been 20ES/day.  In contrast the VMG1312 with the 52Mb/s speed cap has given 46ES/day on its first day.  From tests I estimate that I could run the HG612 with a cap of 58Mb/s rahter than the current 52Mb/s to have the same error rate. 

   In short in spite of its higher attainable it needs to be capped to sync about 6Mb/s slower than an HG612 to have the similar errors.  I had roughly the same result with the VMG8324 vs an HG612.  If your on a line with lots of error head room this won't matter and you will be happy the ZyXEL's extra attainable.  However if like me you have a noisy line which with an Hg612 is below but close to the maximum for retaining fast path, then with no speed capping the ZyXEL's quickly send you interleaved with their higher error rates.  This result probably depends somewhat on the exact line characteristics but it seems consistent to me.

    For the moment I may leave the VMG1312 in place -- I do like a single unit and unlike more vertical units the VMG1312 fits into a  really neat position.  All of my wireless use is three brick walls separate from the modem and more than  2.4GHz, 802.11 b/g/n just does not make it. the VMG1312 is doing fine in that respect.  The current speed cap to 52 is  rather severe but I honestly just can't tell the difference and when used over wireless there is no difference. At this speed the error rate may be 3 times more but it is not significant.

   I will probably run through all the AAISP recommended settings to whether anything else can be improved.

  @npr under remote mgmt I don't have ICMP protocol as an option, maybe that is not the way on a VMG1312 or maybe I need to do something  else first. I will keep my eyes open while reviewing the many GUI settings.

[npr]

It's only recent firmware that has supported that feature in the vmg8924 -- FW L.11 iirc.

A firmware was released in Jan2016 for the VMG1312-B10D, not sure if that's your exact model.
http://www.zyxel.com/euosearch/dl-search.aspx?mci_country=uk&mci_lang=en&keyword=VMG1312&submit=Search

[PhilipD]

Hi

The maximum number of ES per day over a 60 day period at ths speed has been 20ES/day

Are you OCD inclined   20 ES (or 46 ES) a day is absolutely nothing to worry about and it's how ADSL/VDSL is designed to work, it is purposely designed to run with some errors as that is the only way it is possible to push these sorts of data-rates down a cable designed for nothing more than a tinny voice call. If you work that out in percentage terms it is an error rate of 0.00023%, which is nothing, and then consider it probably happened on empty packets (i.e packets that didn't travel with any user data in them) it simply is crazy to worry about it.  You are probably having more error seconds caused by general packet loss then caused by your modem, of course that statistic isn't reported to get worried over.

Why not just let the modem sync at the higher speed and let the equipment manage itself, and stop worrying about error seconds, you only need to worry about them if you can't access the internet or your modem keeps dropping the line

Regards

Phil

[les-70]

 @PhillipD  It is not OCD.  I think I said that 46 is very low.  However as far as going full speed is concerned I have done that.  With an HG612 I go interleaved within a few weeks due an odd bad day.  With a BCM 63168 device like the Zyxel I go interleaved within a few days.   With a speed cap to 64 which is about what interleaving gives me I still go interleaved every few months.  I notice interleaving but not the sync reduction and 52 is a rather over the top reduction that I had simply been testing.  Given my past bad XyXEL experiences it simply seemed a safe configuration to evaluate things.

 I am also with TTB who seem to provision on stable (that is consistent with my events and what they say) and not even standard and certainly not the fast profile.  I keep thinking of changing as I ought then to be able to do as you say.

[PhilipD]

Hi

My error seconds are around 10 to 20 an hour so ~240 to 480 a day and I'm still on fastpath, they do seem to be play it very safe your ISP.

Regards

Phil

[npr]

My ES are around 40 per day, and I still can't get off interleaving, been stuck on it for about a month now!
I'm with Les on this, I cap my sync speed to try a stay off interleaving but mostly to maximise my upstream speed.
I'm on a long, mostly aluminium line, without managing the errors the up stream sync has in the past deteriorated down to 0.5 Mbps. Managing the errors enables me to, mostly, keep a upstream sync around 2Mbps.
ISP Plusnet -- but only for another week.  

[burakkucat]

ISP Plusnet -- but only for another week.  

Then, all being well, to whom will you be migrating? 

[npr]

BT   
With much trepidation I had to break a vow, I made 20 years ago, of never again.