Friday, May 10, 2013

Upgrading to wheezy (Debian 7.0)

I decided to start the upgrade after all. And it looks like I need to do a couple of reboots in the process, so I think I'll just bring Lambda back up when I'm done. ETA 7am (PDT) ?

Update: Well okay, make that 6:10am PDT. We booted successfully with the new kernel. The rest of the upgrade is evidently going to take another 3 hours and since I don't want to be putting you all through withdrawal, the server is back up -- and it's possible I won't actually need to reboot at the end of this anyway, so... enjoy. If we go down again, it'll be around 9-10am, if not then yay.

UpdateUpdate: Did I say 9-10? I meant more like 10-12noon. Sorry. Anyway, it's done; I even got in one more reboot at the end to make sure everything is in a known state, so hopefully we're done with this for a while. @version has been suitably updated.

Saturday, March 23, 2013

Network reconfigured

Ok, the new firewall and new IPs are in place. And the various new IP addresses appear to have propagated through DNS Land. The one beginning with 64 is the one you want.

It may be that some of you are currently connected via the old IP (the 69... one); all this means is that the next time we have a firewall restart and your connection drops there is a remote possibility you'll need to either flush the local DNS cache to grab the new address (ipconfig /flushdns on Windows boxes), or restart your client, or reboot your machine (in which case you might want to consider upgrading to a modern OS) -- but most likely everything that's needed to happen has happened already and a simple reconnect should Just Work.

There's still at least one more server reboot to do once I finish reshuffling the various websites and services. That will probably happen later this week.

Wednesday, March 13, 2013

IP change coming

Owing to a number of factors (**), I am completely overhauling my network for the first time in 8 years (really 12 years -- yow, have I been doing this that long? -- though the changes I did when I switched away from Covad/IINet back in 2004 were relatively minor compared to what I'm doing now).

The main user-visible thing is that various server IPs, including LambdaMOO's are going to change and, at the very least, your connection will drop when I throw the switch. There will also be some number of minutes of downtime while I futz with the cables or if I have to do unexpected extra debugging on the new firewall. Once that's done, assuming a just world, your DNS should be keeping up with the changes, so that all you'll have to do is reconnect and things will Just Work.

Admittedly, if I manage to do this without taking the server down, I may feel compelled to do a server reboot some time afterwards just to make sure things will continue working the next time the power goes out. We'll see how this goes.

(**) Fafner, my vintage-1998 Pentium box that boots off of a floppy that I use for a firewall is getting harder to maintain now that I no longer have any other working machines that possess a floppy drive. And with djbdns no longer being maintained and not keeping up with the various load-balancing/cloud-whatever-bullshit games that akamai and friends are playing, I need a new DNS server. And I also needed more IP addresses, which Megapath (nee Speakeasy) appears to have finally come through with --- only took them a month to figure out how to do it; evidently, the post-merger layoff of the Speakeasy people took place a little too quickly and there are now various chunks of network infrastructure they don't fully understand anymore and I'm on one of them. bleah -- and because what I'm getting is a completely new block that has to be routed via my current address, I have to completely redo how my DMZ works. So...

Friday, December 21, 2012

another power outage

Our power went out roughly 10 minutes ago. No ETA for restoration (we were one of the first people to call it in, so...).

The weather here is, of course, perfectly clear (apart from the small matter of the sun having set already; stupid solstice...)

Update 3:39am Sat:  and back up 6 hours ago...

Thursday, September 20, 2012

power outage

so we lost power at around 2:15 for slightly less than an hour. Looks like everything came back ok, so...

Monday, July 09, 2012

seeing lightning

... and now it's starting to rain. we'll see.

Sunday, June 03, 2012

power outage

So at around 7:30 this morning, I awoke to the sound of the machines beeping and spluttering (wotan's sound system was still hooked up even though I haven't booted that machine for months... makes a really loud popping every time something happens with the power, and this time it was a bit more like a fireworks show...) and the house lights flickering like crazy. hagen and alberich had apparently already been through multiple reboots by the time I got to the power strips to shut them off. And then things finally went dead.

Power is still out (took me a while to remember that emma's iPad gets G3 service, making possible this msg). The weather is perfectly clear and the pse website has no service alert posted, meaning this is an entirely local affair, probably just limited to our neighborhood, probably just a (really nasty) equipment failure from the looks of it.

I have a bad feeling about this one, but we'll see...

Update (11:20am):  And we are back.

The last mail server log entry was at 6:54am, and since we get a pretty continuous stream of spam these days, that's a pretty reliable indicator of when the problems first started, meaning our power had been doing that on-and-off-flicker shit for a full 40 minutes before I woke up. Apparently we never actually made it through a full reboot, just kept getting whacked in the middle of fscking a 500G disk.

I love ext3 filesystems. No, seriously. Just in case you were wondering.