This morning I was watching the approaching lightning storm at
blitzortung.org getting closer. There were various strikes in Ireland, in South Wales and then up here in the Small Isles there was a strike in the sea just north of Eilean Chanaidh which is just a few miles to the west of Skye. I should have pulled the modems at that point but, stupidly, I didn’t and before I knew it
there was a strike right on the Heasta road a few hundred metres north of the house.Janet pulled all the modems and the Firebrick went into failover mode, switching to internet access over 3G using its 3G USB ‘dongle’ NIC. This promptly racked up a bill of £13.68 for AA/AQL/Three 3G SIM network charges. AA said that 0.682Gb had been transferred and I’m presuming that that means
bits, not bytes, because of the lower case “
b” even though such ”bits“ usage seems
really unusual. I make that 85.25 MB (bytes). This huge amount was lessened by the fact that I later connected my iPad to Janet’s iPhone, which has an EE SIM and which was sharing out its EE 4G network connection over 802.11 (a sharing-over-802.11 facility which Apple calls ‘hotspot’). Janet’s EE account has a lot of unused ‘data’ allowance which I too could take advantage of on my iPad by using Janet’s ‘hotspot’ service.
When I later dared to plug the modems back in, it then looked as if all four lines (apparently) were down. A bit of diagnosis, then trying a new known good modem in line 1, and things still didn’t work, and it turned out the lightning had completely
destroyed the 8-port ZyXEL VLAN/MUX switch. The copper DSL lines themselves though seem to have all survived without damage.
Janet then plugged in the
spare/backup VLAN MUX switch that we had just had delivered a week or so ago from Andrews and Arnold, and ta-
DA! Two DSL lines came back up.
Talk about timely, so lucky that for no particular reason I had decided to be prepared and recognise that the only piece of hardware for which I did not have on-site swap-out kit was the small VLAN/MUX switch. I even have a second firebrick, an FB2500, but I realised a few weeks ago that the small switch was the weak link, because I had no spares. And the spare was delivered just in time for the killer lightning strike.
The next job this afternoon was to configure two new spare modems from the reserve stock to go into those slots. I spent what felt like an hour faffing around checking the contents of config files because I hadn’t left them annotated properly. I was in pain and over-tired and was sitting in a chair in the office helping Janet with all the plugging and unplugging. I just ended up
too tired, couldn’t think straight and in too much pain so I actually couldn’t configure the modems. I realised I would have to do a pin-in-hole Johnson/Burakkucat reset first, in order to gain control because the new modems I had obtained from eBay were not talking to me, so a full reset was required before I could even start configuring them. I was just too tired so decided to abandon the work of modem-configuring until the next day.
So in the end I plugged two good, already-configured DLink DSL-320B-Z1 modems into slots #3 and #4 and then all four lines were good for the night.
Tomorrow I will have to attack a load of modems,
at least two, poking them and configuring them; not just two, I should do the backlog of modems I have bought but have never configured.
The small switch was not showing per-port LEDs above the ethernet sockets but the "Power" LED was lit, so it seems certain that that MUX switch is indeed really stuffed. Could anyone tell me how a switch decides to light one of these port LEDs, what criterion it uses?
The replacement small switch shows lit LEDs above the ethernet ports that have something plugged into them, I don’t know if that means something at the near end, far end or if both ends have to be ‘good’ in some sense.
- Lesson to be learned - lightning can appear to ‘jump’ >20 miles with no warning. You see a strike on the lightning maps that is some distance away and you think ‘oh the danger zone is over there’ which is completely false; the thundercloud could already be right over the top of you, and when you’re looking at the lightning strike maps, nothing is going to tell you where the undischarged dangerous clouds ( ie those about to discharge ) are at the present moment. I have never seen a situation like this before.
- Minimum radius. I will have to be much much more paranoid in future. If there is a strike anywhere on the lightning map within say 70 miles radius then it’s time to unplug the modems.
- I really need to have modems ready configured, standing by, and that means ready to go, not needing configuration first. When I get new modems for use in reserve stock, I should be resetting them and configuring them straight away. It also means that newly arriving modems then get inspected and can get tested to some very basic level.
- And I also need to document everything, there’s now a program that I’ve written for the iPad that writes slot-n adapted versions of a master template modem config XML config file, but it isn’t at all obvious to the reader how to declare/define a new / updated master template config that is used by the slot-n copy/adaptation program. Firstly I could not just see where the master config is set, where it’s definition lives and secondly I couldn’t see which file is the basis of this because there was no documentation. Grrrr. :-(