O Canada! > Canadian > TekSavvy > [Cable] Toronto Outage Check-ins

reply to Window

Re: [Cable] Toronto Outage Check-ins

reply to Window
back up since this morning at bathurst and st clair.
but it wont allow me to release the address for another device,
so had to resort to cloning the mac address.

reply to geneDT

reply to Window
Queen E. and Carlaw, out since 2:30 am exactly on Dec. 21.

reply to InvalidError

Tracing route to google.ca [74.125.226.82]
over a maximum of 30 hops:
 
  1    <1 ms    <1 ms    <1 ms  192.168.1.1
  2     9 ms     7 ms     7 ms  10.124.0.1
  3    20 ms    18 ms    19 ms  69.63.242.25
  4     7 ms     7 ms     7 ms  fallowfield1.cable.teksavvy.com [24.246.55.22]
  5    24 ms    13 ms    13 ms  fallowfield1.cable.teksavvy.com [24.246.55.17]
  6    14 ms    15 ms    14 ms  2120.ae0.bdr01.tor.packetflow.ca [69.196.136.66]
  7    28 ms    13 ms    13 ms  gw-google.torontointernetxchange.net [206.108.34.6]
  8    14 ms    13 ms    13 ms  216.239.47.114
  9    16 ms    14 ms    14 ms  64.233.175.98
 10    13 ms    13 ms    14 ms  yyz06s07-in-f18.1e100.net [74.125.226.82]
 
Trace complete.
 

Oh the confusion over what a node and a node split is and how things must work at the head end ...

As I understand it, this is how this works. I am not a good enough artist to produce a pretty picture and don't have very good tools to do it either!

A node is (historically speaking) where cable segments (the cable that you're physically connected to usually at the street) join together and are mixed together onto fibres for the trip to the head end.

So (again historically) a node split was where some of the cable segments are split out of an existing node, and are moved to another cable to fibre convertor and sent up a different fibre.

A node split is done so that fewer customers are connected to a single fibre so effectively there are fewer customers sharing the bandwidth available.

Now, today, these "physical" nodes have been/are being upgraded so that far fewer segments are combined into one fibre so more fibres are "lit". To ensure that CMTS equipment is more efficiently used, the fibres from the cable segments are combined at the headend to form "virtual nodes". This means that say there's 40 customers in one neighbourhood on a segment, and 40 in a totally different neighbourhood, these two can be combined into a virtual node and fed to the CMTS.

So, there can be "physical node" splits, which requires truck rolls and new equipment, and there can be "virtual node" splits, where different segments are split and combined virtually by control from the head end.

Notice that cable segments are not separated into Rogers customers and TekSavvy customers. There is no identification of Rogers or TekSavvy customers at a cable level. That is all done at an TCP/IP level. And at a cable segment or node level, that all knows nothing about the TCP/IP network.

When nodes are split (physically or virtually), it may mean that customers are moved from one CMTS unit to another, and in particular to new CMTS units.

Now what I understand to be happening is that Rogers identifies Internet DHCP requests (i.e. those for your system(s) as opposed to the modem DHCP requests) differently for TPIA customers like TekSavvy from requests for their own customers and ties them to CMTS units, so when customers get split off onto a new CMTS unit, the DHCP server doesn't recognize the TPIA CMTS requests. It seems that they have to manually "move" the recognition of all the TPIA DHCP requests to service the new CMTS unit. Not being any too clever, it seems they don't have an automated process to do this, so they have to identify all the TPIA customers they messed up by the node splits to create a move process.

It's almost as if they wait for the complaints to find out who they screwed over!

On the IP network, the TekSavvy traffic is separated off at the first router from the CMTS (which I refer to as the CMTS router) and sent to the first TekSavvy router at the head end (this and the second TekSavvy router at the head end form the "POI"). Node splits may send your traffic out to a different router, but that only happens to traffic AFTER you get your system IP address. Rogers and TPIA DHCP requests are filtered off to the DHCP servers at the CMTS router. DHCP requests don't go to the TPIA's POI routers at all, so it has nothing whatever to do with the TCP/IP part of the network.

I think Marc or Gabe explained that they have more than double the IP block capacity on all POI's, so even if Rogers rerouted some customers to other POI's it shouldn't have mattered.

Remember that TSI had to wait a long time to get enough IP's from ARIN before they could expand beyond the 6 POI's in the GTA. So they've got a lot.
--
GO LEAFS GO!

reply to Window
The problem apparently is that Rogers not only allocates on a per head end POI basis, but on a CMTS basis. We've noted that at a given head end, there is now 1 POI router per POI connection to TekSavvy. They are apparently simply not moving the available IPs to cover new CMTS units. Now it may be that the POI router is fed from selected CMTS units which is also causing some division of IPs problems, but the bottom line is the screwy way they are allocating IPs.

reply to Window
Well, out since Dec 22. I called to check on my ticket. Rogers note was to check another computer as they thought maybe the network card in my laptop was the fault. Of course we checked both computers when I had the ticket made and the router but Rogers reply was to check again. We did, of course this didn't change anything. Unfortunately had to cancel with TSI, called the evil phone empire and they are coming install Fibre 25 on thursday. Wish I didn't have to cancel but I work from home and need internet service.

Funny thing is there is no phone message at TSI about general outages in Toronto anymore. I guess every thing is all cleared up, except for the whole pesky no internet thing.

Poor guy at TSI in tech support was pretty frustrated about the whole thing, and you could tell he had been dealing with a lot of this crap recently.

reply to Window
Checking in from Jarvis/Adelaide. Been down since the 22nd. Was told they ran out of IP addresses and to keep power cycling the modem until more IP addresses are available. At this point my modem will probably get fried before the Internet's back up.

reply to Window
Still out since the 18th in Southwest Etobicoke, Sherway Gardens Area.

reply to Window
Out since the 22nd at Sherbourne and Front.

reply to sbrook