A decade ago, Joyent accidentally rebooted an entire datacenter, an experience that I described in my 2017 GOTO Chicago talk, "Debugging Under Fire":

https://www.youtube.com/watch?v=30jNsCVLpAE

On the next Oxide and Friends, @ahl we will be joined by folks who were at Joyent a decade ago, both to recall the fateful outage and to reflect on its ramifications, both at Joyent and beyond. Join us, Monday, 5p Pacific:

https://discord.gg/dqUCRwsx?event=1243638578484088842

@bcantrill Ah right, that was the talk I was thinking about when I read this last week: https://blog.danslimmon.com/2024/05/15/ask-questions-first-shoot-later/
Ask questions first, shoot later

The fact that fixing and diagnosing often converge to the same actions doesn’t change the fact that these two concurrent activities have different goals. The goal of fixing is to bring the sy…

Dan Slimmon
@bcantrill I remember that incident Jabber discussion! And do I miss the bots!
@mamash @bcantrill I’ve either completely forgotten this (which would be typical) or it was after my time. Will enjoy the story telling regardless! :)
@mamash @bcantrill I do remember the time we were working so long to hit some deadline that we “lost” Ben on the conference call for a few minutes and when he came back on and David asked what happened all he said was “I can’t see”. 😂😳
@mmayo @mamash @bcantrill oh I very much remember this event and the jabber chat! I was logged into the headnode at the time and suddenly it started rebooting…
@bcantrill this promoted me to listen to more of your monktober fest talks. Now I have coffee on my work keyboard from laughing.

Great talk. In case anybody might think the Gimli glider was a single type event: https://en.wikipedia.org/wiki/List_of_airline_flights_that_required_gliding

And wanting to make sure Canada remains a leader here In 2001 Air Transat 236 from Canada ran out of fuel due to shockingly bad maintenance decisions and very poor flight crew diagnosis of the problem.

https://www.youtube.com/watch?v=tJLTxFOxj4s

List of airline flights that required gliding - Wikipedia

Your description of a B767 as "a brick" is a bit off. The B767 has a L/D max (best case glide ratio) of ~20:1. The aircraft is designed to glide and be landable if needed. That does not take away from the wonderful job the flight crew did here to get that aircraft on the ground.

B767 and other airliners actually have better glide ratios than many light aircraft. But things are greatly harder due to the high energy, and very complex systems involved...

... many single engine light aircraft have L/D max as low as ~ 7:1-8:1. It is not hard to fly those, it's done by pilots all the time in simulated engine failures during training. I'd not describe those flying like a brick either. (Interestingly the glide ratio of many light aircraft is significantly degraded by the propeller windmilling).

Again, not to take away from the skills here but the B767 was not a brick. It likely helped that the Gimli captain was an experienced glider pilot.

@bcantrill @ahl It feels good to get back into more traditional oxide spaces from before what seems to be a massive disruption around the launch and support of the rack
@bcantrill @ahl oh i enjoyed watching that, thanks

@bcantrill Thanks for a great episode! I got interested in how you implemented loading of the host OS, to see if I should implement something similar in my small homelab.

Found RFD 284, which refer to RFD 281 in the background section. From that summary, it looks like 281 is what I'm looking for?

@bcantrill You liked my post, so I guess RFD 281 is correct. Would it be possible to make it public? Or are there some NDA's that are stopping you?